NV Digital Open Operator System (NVoOS)

Version: 1.1.71 Release Date: 2026-09-05

See § Previous Releases for all version history.

🆕 v1.1.71 Highlights: A rate-limit, model-catalog & ecosystem release. REST rate-limit blocks are now liftable — the general request limiter uses fixed-window accounting (honest retry_after, no infinite slide) and flags the restriction registry via a new wp_mcp_ai_rest_request_rate_limit_exceeded action, so blocked users appear in the Command Center Restrictions tab with the existing Lift button. wake_up_context enforces wing/room scope (Graphify graph anchors no longer leak cross-wing memories into a wing-scoped block). The September 2026 model catalog refresh ships — 228 models (gpt-5.6 family, gpt-6-astra preview, gpt-image-2, claude-opus-5, gemini-3.6/3.7/3.8-flash, kimi-k3), retired DeepSeek/gemini-3.1-flash/imagen-4 IDs with migration-map successors, pricing drift fixes, and new defaults (gemini-3.6-flash, gpt-image-2, kimi-k3). The checkout-api addon joins the standard build + test pipeline and its token/crypto classes now derive salts from wp_salt() (no more latent fatal on installs without salt constants). The Content Graph ecosystem advances: the standalone plugin ships its 1.0.4 visual experience system, Content Graph AI gains the assistant-builder block set + settings shell, and the platform completes its Wave E2 queue layer (async job queue → queue manager → job queue manager → dead-letter queue). Connectors links point at the real core screen. Coding-time agent skills: 54 (new mcp-ai-wpoos-updates skill). Tool count: ~303 base + ~1,263 Pro (~1,566 total).

MCP Specification: 2026-07-28 (Stateless Core, Full Compliance)
Maintained by NV Digital
License: GPLv3 or later
Requires: WordPress 6.0+, PHP 7.4+
Patent Status: Patent Pending (Application #19/410,504)
Documentation: Grade A (95/100) — 1,617 files across 12 directories, 108 admin screenshots, 100% feature coverage

🔍 For Reviewers & Auditors

New to this repo? Start here → docs/project/FOR_REVIEWERS.md

That document answers every common question in one place: what the project is, current security posture, what's production vs experimental, PHP version requirements, AI development methodology, compliance status, and scoping advice for a limited-budget review.

Quick links for reviewers:

• Addon Inventory — what each of 27 addons does and its status

• Security Posture — current state of all 50 audit findings

• Compliance Traceability — every .org rejection reason → commit → verification command

• AI-Assisted Development — methodology, transparency, and what to scrutinize

• Architecture Overview — component diagram and data flow

Related MCP server: WP MCP Ultimate

📑 Table of Contents

Getting Started

• 🧩 Overview

• 🎯 Our Mission

• 🛡️ Active Security Monitoring

• ⚠️ Warranty & Safe Use

• 🏗 System Architecture

• 🚀 Features

• 📦 Installation

  • 🌱 Beginner 3-Step Install (Try it on Your PC)

• 🔌 What You Lose Without Third-Party Plugins

• ⚙️ Configuration Checklist

• 📚 Documentation

Core Functionality

• 🧠 Memory & Tool Stack Overview

• 🛠 Built-in tools & automations

• 🗨️ Front-end chat surfaces

• 💬 Frontend Shortcode

Addons & Extensions

• 🏗 System Architecture (covers addon ecosystem)

Orchestration & AI Features

• 🏗 System Architecture (covers orchestration, harnessing, MCP servers, memory bridge)

AI Providers & Integration

• 🧠 Language Model Providers

• 🧱 ChatKit Integration

• 🌐 Crawl4AI Integration

• 📡 Job Notification System

• 🧊 Elementor Widgets

Performance & Optimization

• ⚡ Message Bundling

• 🎯 Agentic Loop Token Management

• 🔄 Chat Performance Optimizations

• 🌐 Mesh Compute Routing

• 🔗 Federation & Discovery System

Remote MCP Setup

• 🔒 MCP Server Authentication

• 🌐 Connecting Remote MCP Clients

• 🛰 REST API Endpoints

• 🌊 SSE Streaming Support

• 📝 MCP JSON-RPC 2.0 Endpoint

• 🔑 Assistant API Credentials

• 🎫 Token Management UI

Assistant Management

• 🛠 Assistant Editor Overview

• 📊 Assistant Storage: CPT vs CCT

• ⚡ Assistant Tool Shortcuts

• 🧠 Agent Skills

• 👔 Professional & Team Layers

• 🧵 REST Chat Payloads & Attachments

Development

• 🐳 Local Development with Docker

• 🧑‍💻 Development Tooling

• 📦 NPM Packages

• 🧪 Testing & QA

• 🧩 Hooks & Filters

• 🧰 WP-CLI Commands

Reference

• 🔐 JetEngine Capability Reference

• 🛰 JetEngine REST API Reference

• 🧮 Usage Tracking

• 🧷 Attachment MIME Controls

• 🧾 Logging

• 🧾 JetEngine REST Endpoint Report Helper

• 🔌 Optional Tools & Dependencies

• ✅ Manual QA Scenarios

🗺 Repository Map

🧩 Overview

Real-time AI Orchestration Toolkit for Wordpress - NV oOS is a modular AI framework (Object-Oriented System) for WordPress that connects your site's data with 15 language-model providers: OpenAI, Gemini, Anthropic, DeepSeek, OpenRouter, Baseten, Kimi (Moonshot), Z.AI (GLM), DigitalOcean, NVIDIA NIM, Cloudflare Worker AI, Ollama, LM Studio, Hugging Face, and Flowhub. It allows you to create and manage AI Assistants that can interact with users, access WordPress data, and perform custom tool functions.

✨ What's New at a Glance (v1.1.71)

• 🔓 REST Rate-Limit Unlock & Fixed-Window Timer (PR #6322). Users blocked by the general request limiter never appeared in the Command Center Restrictions tab, so admins couldn't lift the block. Two bugs: the limiter never flagged the restriction registry, and the window was sliding (every request rewrote the transient with a fresh full-window TTL, so steady traffic kept the block alive and retry_after always reported the full window). check_rate_limit() now uses fixed-window accounting ({count, first_seen}, TTL never extended past window end) with honest remaining-time retry_after; a new wp_mcp_ai_rest_request_rate_limit_exceeded action makes WP_MCP_AI_Restriction_Registry flag a rate_limit restriction (auto-release at window end) — the block shows in the Restrictions tab + Token Manager with the Lift button, and lifting clears the request window. Guest (IP-keyed) blocks expire on their own.

• 🧠 MemPalace Wing-Scope Enforcement (PR #6327). WP_MCP_AI_Tool_Wake_Up_Context::matches_wake_filters() now enforces wing/room exclusions — the Graphify graph anchors only boost scores and never excluded out-of-scope memories, so wake_up_context in auto mode leaked cross-wing memories into a wing-scoped block once the Graphify addon loaded. The PR also ships four test-isolation fixes (transcript mock restore, shared registry teardown, provider-key isolation, logging cache reset).

• 🛒 Checkout API Into the Addon Pipeline (PR #6315). bin/build-addon-zips.sh now produces build/nvoos-checkout-api-v<version>.zip, the addon's 6 suites join phpunit.xml.dist, and the build workflow gains a tests job. The new suite exposed a latent production fatal: the token/crypto classes read AUTH_KEY . SECURE_AUTH_KEY directly — now wp_salt( 'auth' ) . wp_salt( 'secure_auth' ) (identical on normal installs, stable fallback otherwise).

• 🧭 September 2026 Model Catalog Refresh (PR #6328). 217 → 228 models: new gpt-5.6-sol/-terra/-luna, gpt-6-astra preview, gpt-image-2, claude-opus-5, claude-fable-5.1, claude-mythos-5, gemini-3.6/3.7/3.8-flash, gemini-3.5-flash-lite, deepseek-v4-flash-vision-exp, kimi-k3, kimi-k2.7-code; retired DeepSeek chat/reasoner/coder + gemini-3.1-flash + imagen-4 with migration-map successors; pricing drift fixes (DeepSeek v4-flash input 10×, Claude Opus inputs $15 → $5); new defaults (gemini-3.6-flash, gpt-image-2, kimi-k3); ~45 derived files updated; 517 model tests / 8,290 assertions green.

• 🌐 Content Graph Ecosystem (PRs #6316–#6321, #6325). The standalone nvoos-content-graph plugin ships its 1.0.4 visual experience system — theme engine (PHP/JS byte-identical, WCAG 2.2 SC 1.4.11 ≥3:1 palettes), appearance settings tab, 24 SVG type glyphs, legend/minimap/zoom/keyboard explorer chrome, GET /edges route, export at 1–3×, block/shortcode attribute parity, and a checkout fallback redirect (#6318). Content Graph AI gains the assistant-builder block set (selector, tools grid, knowledge base, builder) and a validate-then-sanitize settings shell (#6316, #6317). The platform completes its Wave E2 queue layer: AsyncJobQueue → QueueManager → JobQueueManager (atomic SELECT … FOR UPDATE SKIP LOCKED) → DeadLetterQueue, byte-identical to the base stack (#6319–#6321, #6325).

• 🔗 Connectors Links Fixed (PR #6314). The "Settings → Connectors" links now use admin_url( 'options-connectors.php' ) — the old admin.php?page=connectors target isn't a registered page.

• 🧪 Post-K16 Singles (PR #6327, CI run 91942465749). Order-dependent singles closed: transcript session-key mock restore, shared tool-registry teardown (Pro OKF tools were wiped by clear_tools()), Site Health provider-key isolation, transcript-mining logging cache reset. The mcp-ai-wpoos-test-suite skill grew to 40 patterns.

• 📚 New Coding-Time Skill (PRs #6323, #6324). .agents/skills/mcp-ai-wpoos-updates/ codifies the two maintenance tracks — docs & release catch-up (Track A) and model catalog & config updates (Track B). Coding-time agent skills: 53 → 54.

✨ What's New at a Glance (v1.1.70)

• 🛡️ Shell-Call Hardening on exec-Disabled Hosts (PR #6295). The Pro Packages page (and five other Pro admin pages + the OCR service) fataled on hosts with exec in disable_functions — on PHP 8+ @ cannot suppress the disabled-function Error. Canonical wp_mcp_ai_check_nodejs_available() / wp_mcp_ai_get_nodejs_version() helpers probe exec-first and fall back to WP_MCP_AI_Process_Service (proc_open); the OCR service gains is_cli_tool_available() / run_cli_command() (Windows-aware) and guarded proc_* calls; seven CPT/settings pages clamp negative inputs with max( 0, … ) instead of absint() (absint('-5') → 5).

• 🐛 Wave-5 Production Fixes (PRs #6280–#6312). Transcript attachments display metadata preserved on save + default max_tokens capped by the model's window (#6280); Pro dashboard event stats accept underscore and dashed slugs (#6281); validated web_search profession_tags aligned (#6282); schedule presets drop a phantom validate_contact_data step and put the hook in schedule_data (#6283); mesh peer URLs validated before esc_url_raw() (#6285); opt-in guest memory capture writes under a shared guest bucket (#6286); JetEngine gates require JET_ENGINE_VERSION + a physical CCT-table probe, the JFB submissions tool restores id/flat params, and the token manager uses the cached Pro tool map (#6296, #6300); complexity routing restored — the default model no longer short-circuits select_model() (#6298); Auth0 audience validation is structural, image tools check read_post against the acting user, and user IDs validate as positive ints (#6301, #6305); markup REST errors return 400, the legacy SSE handshake is filterable, skill packs reject numeric-keyed entries and always fire their installed action (#6302); clamp sanitizers + plugins-integration terminate seam (#6303); Elementor is_elementor_editor_init() + static::is_available() (#6304); provider-prefixed model-slug token limits, Ollama native tool-calls fix, published_date preservation, wp_mcp_ai_missing_image (#6306); agentic events reach the recent-activity feed + millisecond trace durations (#6308); K16 bundle — sanitize_key-stable webhook IDs, settings-repository blob fallback for runtime gates, acting-user chart unfiltered_html, social-publish date reset, PSO inflections, site-builder h1–h6, schema-verifier required (#6311); curriculum exporter skips empty structures (#6293).

• 🧪 PHPUnit Repair Campaign, Wave 5 (~29 test PRs, #6280–#6312). Two CI triage runs (CI-TRIAGE-91006542428.md + CI-TRIAGE-91771001271.md) drove the clusters closed: transcripts, charts/dashboards, professions, presets, Cloudflare, mesh, memory, CPT management, PayHere, Mubert, OpenAI Images, composer drift, Content Graph AI, pro schedule AJAX, model routing, Gemini, JetEngine/Graphify stubs, OAuth, REST/skill packs, admin settings, Elementor, permission gates, HTTP clients, orchestration budget, logging, K16 misc singles. The mcp-ai-wpoos-test-suite skill's cluster board is compacted to one completed wave-5 block (37 patterns unchanged).

✨ What's New at a Glance (v1.1.69)

• 👁️ Vision Analysis Toolkit (Pro, PR #6267). analyze_image_objects detects and counts objects per category — detection (HuggingFace OWLv2 / Ollama vision), vlm (OpenAI / Anthropic / Gemini JSON-enforced counting), and hybrid (detector owns counts, VLM only renames mislabeled categories) — with annotate=true returning a GD box-annotated copy as a media attachment. Gated by a new NV oOS → Vision Analysis settings page (enable_vision_analysis_toolkit, off by default) with SSRF URL-guards. See docs/toolkits/vision-analysis-toolkit.md + docs/proposals/vision-analysis-object-counting-tool.md. Pro tool count +1.

• 🐛 tagDiv Admin Compat & Metabox Crash Fix (PRs #6266, #6278). The four SiteKit tools no longer return the non-existent CAPABILITY_CAN_USE_IF_ADMIN constant (string flag arrays; the assistant tools metabox crashed exactly at sitekit_get_adsense on nugl.com). The sortable shim prints a bundled jQuery UI 1.14.2 Sortable inline whenever td_wp_admin is enqueued (print-queue detection); the WordPress media chain is forced as direct head tags from core-registered handles; tools metabox CSS is enqueued in the head.

• 🔧 DeepSeek Empty-Schema 400 Fix (PR #6272). Argument-less tools now emit properties: {} (never []) across 29 files; LegacyToolAdapter preserves object maps and upgrades empty arrays; the AI Tool Builder scaffold emits new stdClass() for parameterless tools.

• 🔒 ZipSlip Guard Revived (PR #6270). ZipArchive::$num_files doesn't exist on PHP 8.x, so the guard loop never ran — unsafe archives imported successfully. count( $zip ) now drives all five entry loops (OKF bundle manager + four Pro admin pages), restoring path-traversal rejection.

• 🐛 Fixes. Rate limiter classified by the dispatching request's real HTTP verb with a separate nefarious-monitor counter and dual-shape chat hook tolerance (#6265); one broken tool's shortcut tasks can't fatal the assistant edit screen (#6271); model catalog: gpt-4o 128k context, gemini-2.0-flash video-capable typo, claude-sonnet-4-6/gpt-4o active status, two Qwen entries (#6274); settings save no longer leaves object-cache additions suspended (WP 7.1 inline-async tick locks, #6277).

• 🧪 PHPUnit Repair Campaign Continuation (~9 test PRs, #6260–#6276). Fourth wave: count-tokens params, continuation seam, SSE handler, deterministic cache tests, coverage manifest, incorrect-usage notices, admin registration, subtab sanitizers, 14 Veo suites. The mcp-ai-wpoos-test-suite skill now distills 37 root-cause patterns.

✨ What's New at a Glance (v1.1.68)

• 🧩 Pro SPA v2 Shortcode & Embedded Mode (Pro, PR #6256). New [nvoos_pro_spa] shortcode mounts the Pro SPA v2 in chat-first embedded mode — AgentPanel + drawers, tool shortcuts, OKF drawer, router-free, no admin routes — with per-instance data-config attributes and optional guest mode via the base guest-token machinery. New WP_MCP_AI_Pro_SPA_Shortcode + Pro_SPA_Config + module-registry entry + spa-v2 block. See docs/project/proposals/033-pro-spa-v2-shortcode-proposal.md.

• 🖥️ Hermes Dashboard Fleet Extensions (PR #6249). New top-level extensions/ tree: a fleet monitoring + control plane extension (nv-oos-fleet — Python plugin API over ~35 sites), plus backup-download, external-app-tab, and mcp-tool-shortcuts WebUI extensions with installer + smoke tests. Plans: docs/developer/integration/hermes-dashboard-extensions-plan.md.

• 🔄 Chat-Bubble Stale-Nonce Self-Heal (PR #6225). GET /mcp-ai/v1/session/nonce returns a fresh session-bound wp_rest nonce minted from the request's own auth cookie (no-cache); the chat client silently retries 403 rest_cookie_invalid_nonce failures from full-page caching or session-token rotation on long-lived SPA pages.

• 📚 Docs Hub 0.4.2 (PRs #6246, #6253). Internal links on local pages resolve to #/slug hash routes; TOC anchors mirror github-slugger exactly; "Accept fix" suggestions are directory-relative and ../ links are validated with containment checks; skipped rows show reasons; sync failures surface "Atomic swap failed"; the emoji loader no longer crashes the React SPA. Addon bump 0.4.1 → 0.4.2.

• 🔌 Providers Disabled by Default on Fresh Installs (PR #6255). OpenAI/Anthropic/Gemini default to disabled; dropdowns list only enabled + credentialed providers; the onboarding wizard auto-enables the provider whose key you enter.

• 🐛 Fixes. Google Calendar granted-scopes %20 corruption (#6226); Tiptap mergeAttributes prototype pollution — canvas/document editor pin tiptap 3.30.4 (#6250); fast-uri >=4.1.4 across npm trees (#6244); assistant untrash restores the pre-trash status (#6239); presets gain missing Google/Gmail/Drive/OKF/git tools (#6242); wp_mcp_ai_seed_task_templates AJAX wired (#6243); jQuery UI sortable shim for post screens (#6258); agent-identity canonical-ID int cast (#6232); token-budget catalog dedup + restored wp_mcp_ai_model_tpm_limit filter seam (#6233).

• 🧪 PHPUnit Repair Campaign Continuation (~21 PRs, #6224–#6257, + #6259 merged upstream). A third wave kept the suite aligned with current contracts; the production fixes it carried are grouped above (including the late-merged wp_mcp_ai_attachment_segment_provider filter — providers without a remote file API, e.g. Ollama, resolve attachments to local references).

• 🧹 Housekeeping. Stale 1.1.67 build ZIPs removed (27 files).

✨ What's New at a Glance (v1.1.67)

• 🧩 Content Graph AI Platform v2.0.0 Standalone (PR #6123). The extracted nvoos-content-graph-ai-platform addon now carries its own business logic — Waves A–C + Blueprints deliver the namespace-bridged admin UI (PlatformDashboard, PlatformSettingsRegistry), skill/slash-command/agent bridges, harness router, a 74-skill bundled-skills pack with third-party notices, a knowledge base, and a dedicated PHPUnit matrix. See docs/project/plans/content-graph-platform-extraction-plan.md.

• 🧱 Base+Pro → Content Graph Ecosystem Port, Wave D + D-UI (PR #6142). The AI runtime lands in nvoos-content-graph-ai: chat core (optimizer, caches, rate limiter, summarizer, threads, attachments, transcripts, ChatKit), providers (Zai, Google Maps, Realtime ×3, RabbitMQ, STDIO, file services), model management + analytics/token tracking, security guards, and D-UI assistant pages + blocks/widgets/guest tokens/memory/CLI/MCP JSON-RPC. Content Graph AI bumps to 1.0.4. Tracked in docs/project/ecosystem-port-tracker.md.

• 📧 Google Workspace Gmail + Drive Read Tools (Pro, PRs #6151–#6152). Six new Pro tools — get_gmail_message, get_gmail_thread, list_gmail_connections, modify_gmail_message (destructive-ops gated) and get_drive_file, list_drive_connections — backed by new WP_MCP_AI_Pro_Gmail_Client / WP_MCP_AI_Pro_Google_Drive_Client clients on the shared includes/google/ foundation. Pro tool count +6.

• 🧪 PHPUnit Repair Campaign Continuation (~100 PRs, #6114–#6208, #6209–#6222). A second cluster wave kept the suite green through the extraction/port work and PHPUnit 11 / WP 7.1 drift; the production fixes it carried are grouped in the changelog (crawler contract hardening, REST guards, LLM sanitization delegation, URL-encoded external-API queries, Veo 5s floor, settings/cache fixes, WhatsApp webhook signature rejection, HTTP 304 cached-body re-serve, Veo async-job completion ordering, token-budget display, Shopify client-availability filter seam, Paper Store array-field queries). The mcp-ai-wpoos-test-suite skill now distills 26 root-cause patterns.

• 🧹 Housekeeping. Stale 1.1.66 build ZIPs removed; re-created 1.1.65 ZIPs re-removed.

✨ What's New at a Glance (v1.1.66)

• 🧪 PHPUnit Suite Repair Campaign (~100 PRs). The single-process suite was repaired cluster-by-cluster — REST endpoints, AJAX handlers, provider/client suites, admin pages, chat/channel integrations, CRM, professions/teams, multi-agent orchestration, cron/notifier, federation, healthcare interop, security, logger, transcripts. Most PRs are test-only; the production fixes they carried are grouped in the bullets below. (PRs #6008, #6009, #6012–#6107)

• 🔐 REST, Auth & Permissions. Assistant-access caching — new wp_mcp_ai_assistant_access_cache_enabled filter plus WP_Error caching like successful lookups (#6008); attachment-segment validation errors return explicit 400s (#6009); token-tier endpoint + tier-change audit logging (#6018); permission-callback allowlist refresh (#6019); bearer-auth context sync (#6014, #6016); REST analytics endpoints (#6015).

• ⏱️ Job Queue, Cron & Notifier. Closure serialization fixed for legacy option storage (#6020); custom-table query guards against missing schema (Graphify DB, job store, tenant DB) end the 1.5K-line CI error-log flood (#6022); restored notifier update_status() with dot-preserving job IDs + owner-scoped REST routes (#6036, #6037); job status promoted to running on progress updates (#6039); Little's-law metrics (#6038).

• 🤖 Provider, Tool & Envelope Fixes. Web-search result building restored for Exa/Perplexity (#6042); auto-categorize router/client (#6064); Cloudways analytics adapter auth check (#6062); Media Toolkit tools normalized to WP_Error (#6057); Remove Background path guard + source resolution (#6072); memory-capture failure envelope restored (#6049); A2A webhook logging + delegate tool errors (#6055).

• 🖥️ Admin, Render & Output. Output-buffer leaks fixed across render/AJAX paths (#6045); Graphic Editor Plus image access on WP 6.9+ (#6069); workflow-editor AJAX WP_Error/hook drift (#6070); quiz + document-template admin pages wired (#6089, #6090).

• 👔 Professions, Federation & Addons. Profession meta clamping + primary-role flow, team CPT sanitization (#6043, #6105); federation mesh-sync init (#6044); defensive Graphify admin loading on the PM submenu (#6106); Content Graph AI bumped to 1.0.3 with a permission-check fix (#6056).

• 📚 New Coding-Time Skill. mcp-ai-wpoos-test-suite — Docker commands, CI-log triage, 16 recurring root-cause patterns, and the cluster-PR conventions; coding-time skills now 53. (PR #6108)

• 💳 Checkout API Addon + Content Graph Paid Checkout. New addons/checkout-api/ (v0.1.0, vendor-server only) — Stripe session/verify endpoints, encrypted secrets, custom-table license store, signed expiring ZIP downloads, webhook license issuance/revocation, per-IP rate limiting — plus the paid-checkout client in the free nvoos-content-graph plugin (Stripe Payment Element, verify, license, one-flow install; no Stripe keys in the plugin). Addon count: 27. (PR #6063)

✨ What's New at a Glance (v1.1.65)

• 🧠 OpenAI Reasoning-Model Parameter Fix. o-series and gpt-5 models reject max_tokens (and o-series temperature) — lib/core's OpenAiCompatibleClient now strips unsupported parameters per model (applyModelConstraints()) and retries 400 rejections with corrected payloads on both sync and streaming paths. (PR #5985)

• 🕷️ Media Worker Full-Crawl4AI Proxy (031 Phase 3). New env-gated POST /api/crawl/full + GET /api/crawl/full/task/:id routes forward to a CRAWL4AI_FULL_URL deployment — SSRF-validated targets before proxying, token-gated, 503 service_not_configured/502 upstream_unreachable envelopes — plus a strict-path TEMP_ROOT allowlist (proposal 028 Q5) for Docker shared-volume deployments. Worker stays v3.2.0. (PR #5983)

• 🛡️ Security-Posture Findings Closed (issue #5972). Algorave Tone.js raw eval now requires one explicit confirmation per browser session plus a visible warning banner (capability-scoped); the TMA markdown source map is removed; remaining webhook __return_true permission callbacks carry justification comments. SECURITY_POSTURE.md updated. (PR #5981)

• 💬 Chat, REST & Transcript Hardening. Legacy clients sending a top-level attachments parameter are tolerated instead of hard-400'd; custom message roles work again (wp_mcp_ai_allowed_message_roles); orphaned tool messages are silently discarded before provider dispatch; attachment prep errors propagate to the client; transcript pagination preserves negative values so clamping works; Content Graph AI embeddings stop 500ing and graph context falls back to keyword search without an index. (PRs #5991, #5993, #5994, #5985)

• 🔔 Webhook & Tool Fixes. Google Chat gains a shared-secret verification_token for OIDC-disabled webhooks; Slack link/italic conversion ordering fixed; paper-store import/export tolerate missing collection; scheduled-result block registers idempotently; Graphify treats provider-prefixed *_key fields as sensitive and returns canonical WP_Errors; remote_wp_connection errors direct callers to list_connections first. (PRs #5990, #5986, #6003)

• ⚙️ Slash Commands, Blocks & Admin. Slash-command handler staleness on init re-fire fixed; CSV list parsing accepts "1,2" and "1, 2"; assistant-builder/Pro toolkit blocks register idempotently (WP 7.1 notices); workflow AJAX double-output fix; TPM limits fall back to the bundled model catalog without JetEngine; dangling legacy admin settings removed; WP 7.0 connector registry guard. (PRs #6000, #5997, #6004, #6005, #6006)

• 📦 Content Graph wp.org Refresh. Icons (v5), main screenshot, and a generated page preview for the nvoos-content-graph listing; Contributors field fixed; Content Graph AI chat tester fixed. (PRs #5980, #5988, #5982, #5992)

• 🧪 Test-Suite Cluster Fixes. SSE stream-contract alignment, WP 7.1 bootstrap neutralisation, transcript/slash-command/remote-connection suite follow-ups. (PRs #5995–#5999, #6001, #6002)

✨ What's New at a Glance (v1.1.64)

• 📅 Google Calendar Connection & Shared Google Services. New includes/google/ foundation (OAuth service, Calendar v3 client, scope registry, credential resolver, sync + push) replaces four copy-pasted and drifted Google OAuth flows; a google_calendar connection type lands on both connection surfaces. Six new Pro tools in addons/pro/includes/tools/google-workspace/ — list_google_calendars, list_google_calendar_events, update_google_calendar_event, delete_google_calendar_event, check_google_calendar_availability (freeBusy), quick_add_google_calendar_event — plus a reworked create_google_calendar_event (scope-enforced writes, Meet links) and a sync_google_calendar that actually syncs. See docs/developer/architecture/integrations/google-calendar-connection.md. (PR #5959)

• 🔬 Composio Account Health & Hardening (Pro). A verified account-health engine probes live credentials with catalog-discovered read-only tools (no invented slugs), a new composio_manage_accounts tool handles validate/reconnect/delete/prune, and the Remote Sites Connected Apps table gains a Health column with Verify/Reconnect actions. Auth config resolution, connected-account listings, identity-bound execution, app removal, zero-argument payloads, and proxied provider failures are all fixed. See docs/composio-connect.md. (PRs #5932–#5934, #5936, #5953, #5958)

• 🕵️ Non-Loggable Result Fields. Tools whose results carry capability credentials under innocuous keys (or opaque URL path segments) can declare the exact fields via the new WP_MCP_AI_Tool_Sensitive_Result_Interface; the logger masks them before preview truncation. Logging-only — caller/model payloads are never altered. (PR #5961)

• 📉 Log Hygiene. Per-entry context budgets (fingerprinting + truncation) stop wp_mcp_ai_recent_errors/wp_mcp_ai_recent_activity from ballooning into the megabytes; Data Management gains one-click Compact/Delete for both buffers; credential-bearing URL query params are redacted from every logged string. (PRs #5952, #5954)

• 🐛 Fixes. Validated-tool argument validation restored on Symfony 5.4 (constraint loading was silently skipped — every validated tool accepted invalid input); MCP JSON-RPC -32700/-32600 error envelopes reachable and the diagnostics page re-wired; Pro SPA v2 conversation/assistant sync; vision tools accept a 5–300s timeout; WP_Error envelope drift + cron/memory/elementor tool bugs. (PRs #5957, #5960, #5962, #5964, #5965)

✨ What's New at a Glance (v1.1.63)

• 🧬 Artifact Evolution Phases A–G (Base + Pro). The Continual Harness Evolver and Meta-Harness become a complete, gated, Darwinian evolution loop for skills, prompts, and roles: competing artifact populations with fitness-weighted parent selection, failure-case replay + post-mutation verification, learning-log-aware mutators, a pre-commit admission gate (structural / harmlessness / marginal-gain critics), holdout-gated deployment with shadow A/B serving and drift-triggered rollback, and a cross-cutting governor with a human approval queue and per-artifact lineage graphs. The evolve_harness ↔ Evolver contract is repaired, and every layer is opt-in (default off) with switches in Settings → Orchestration Layer. See docs/project/proposals/007-artifact-evolution.md. (PRs #5923, #5925)

• 🧩 Pro Addons Page (Pro). New registry-driven NV oOS Pro Dashboard → Addons page installs and activates standalone addons whose ZIPs ship in the plugin's build/ directory in one click — nonce + install_plugins + allowlist gated, with non-WordPress components (Media Worker, Cloud Worker, Tenant Router, Schedule Anything SPA) listed read-only. (PR #5924)

• ⚡ Chat Storage Worker Offload. Conversation saves at or above wp_mcp_ai_storage_worker_threshold (default 10,000 chars) offload the expensive JSON.stringify to a browser storage worker; small saves, unload flushes, and worker failures fall back to the synchronous main-thread write, and the threshold filter is the kill switch. See docs/project/proposals/032-chat-web-workers-wiring-implementation-plan.md. (PR #5928)

• 🐛 DeepSeek Empty-Schema Fix. Empty tool-schema property maps now encode as {} (never []) at all three payload boundaries, fixing DeepSeek 400s; legacy tools are wrapped before the first register attempt so the fail-loud registry log only fires for genuinely unregisterable classes. (PR #5926)

• 📚 OKF Skill-Knowledge Conformance. All 91 bundled SKILL.md files gained the OKF v0.2-required type: Skill frontmatter and the four missing cross-linked reference files were restored — the OKF Validate tab no longer reports missing concepts. (PR #5919)

• 🧪 Test-Suite Exit Traps. New bin/sweep-tests.php sweeps every test file in parallel to surface exit-trap and drift failures; AJAX test contracts land in the bootstrap and 24+ test files align with the catchable-die/admin-handler contracts. Minimal production fixes: bare-exit test seams, duplicate CRM workflow-rule class rename, toolkit MCP scope check + nvoos:// resource URIs, SSE/Veo polling-limit filters. (PR #5929)

• 🔧 Pro SPA v2 Slash Commands. /chat, /new, /clear, /copy, /export now insert reliably from the slash-command drawer. (PR #5920)

✨ What's New at a Glance (v1.1.62)

• 📚 OKF Bundle Management (Base). New WP_MCP_AI_OKF_Bundle_Manager gives OKF knowledge bundles a full lifecycle — create, list, rename, archive, delete, ZipSlip-safe ZIP import/export, and health statistics — with the auto-generated skill-knowledge bundle protected from writes. Three new tools (okf_list_bundles, okf_validate_bundle, okf_import_bundle) join the extended okf_write_concept provenance schema (resource/sources/usage_window/verified), and a new OKF Bundle Manager admin screen (Bundles / Browser / Editor / Import-Export / Validate) mirrors the Skill Manager UX. See docs/features/okf-integration.md.

• 🔗 OKF → Skill Bridge (Pro). load_skill now resolves bundle:concept_id names into OKF concepts as loadable skills, gated per assistant (grant metabox, fail-closed), lifecycle (drafts rejected), and an optional minimum trust tier.

• 🤖 OKF Auto-Enrichment & Hybrid Knowledge Router (Pro). The enrichment agent crawls published site content (posts, pages, taxonomy terms) into OKF concepts with cross-links — deterministic and idempotent, AI summaries opt-in via filter — while route_knowledge_query classifies questions across OKF / vector / Paper stores and performs the OKF lookup when it is the primary route.

• 🧭 OKF Skills Drawer in Pro SPA v2 (Pro). A new read-only mcp-ai-pro/v1/okf REST surface (bundles, concept browse/search, assistant skill grants) powers an in-chat OKF drawer: browse bundles and concepts and reference them from the conversation without leaving the chat surface.

• ⚡ Vector Store Tools on the Responses API. OpenAI removes the Assistants API on 2026-08-26 — all vector-store operations now run headerless on the Responses API with file_batches ingestion (bounded polling + single-file fallback); list_vector_store_files gains filter, search_vector_store gains ranking_options. Fixed: 404s on percent-encoded OKF concept routes (%2F now decoded).

✨ What's New at a Glance (v1.1.61)

• 🧠 Agent Identity Bridging in Memory Store & Recall. New WP_MCP_AI_Agent_Identity_Resolver (includes/services/) resolves virtual agent keys (e.g. nvoos-pro-spa-memory-drawer, virtual_planner_1) to the canonical assistant post ID when storing agent context; chat-memory recall merges every alias bucket with per-record stored_under stamps. Memory drawers (base, chat-spa, pro-spa) show wing/room/stored-under chips, an agent-ID diagnostic, a show-all-scopes toggle, and refresh automatically on store events; graph-bridge and scoped-recall failures degrade gracefully. See docs/features/memory/chat-client-integration.md.

• 📚 OKF Skill-Knowledge Bundle Generator. The skill-knowledge OKF bundle is now auto-generated from bundled skills on bootstrap (and refreshed after bundled-skill reinstall) — okf_search and the other OKF tools work out of the box instead of failing with "OKF bundle not found: skill-knowledge". (PR #5911)

• 🔧 undici ^7.29.0 jsdom Compatibility. The undici override is pinned to 7.29.0 — jsdom 29.1.1 breaks on undici 8, which had vitest suites failing in seven addons; all CVE fixes are retained. (PR #5910)

• 📦 nvoos-content-graph wp.org Review Reply & Report shipped and excluded from distribution ZIPs; content-graph CI checksum sync fixed. (PRs #5912, #5906)

✨ What's New at a Glance (v1.1.60)

• 🔒 Restricted-User Flagging & Unblocking. Ephemeral rate-limit and token-budget blocks become persistent, reviewable restriction records with auto-expiry and audit logging — one-click lift on the Token Manager "Restricted Users" panel (Base), a full Restrictions tab on the Pro Command Center, REST/AJAX/WP-CLI operations (wp mcp-ai restrictions list|lift|add), IETF rate-limit response headers, and filterable chat rate limits (wp_mcp_ai_chat_rate_limit / wp_mcp_ai_chat_rate_limit_window). See docs/features/security/user-restrictions.md. (PR #5901)

• 📥 Conversation Import to Transcript CCT (Full, JetEngine). New includes/conversation-import/ subsystem imports ChatGPT, Google Takeout Gemini, Claude, ShareGPT, and OpenAI fine-tuning JSONL exports into the JetEngine ai_chat_transcripts CCT — four new tools (conversation_import_detect|run|status|delete), an admin upload/preview page, WP-CLI, GDPR export/erase, and optional memory mining. See docs/user-guides/conversation-import.md. (PR #5898)

• 🧩 Tool Schema Normalization. Tool argument schemas are normalized before provider payloads (DeepSeek client, REST /tools output, Tool Service, ChatOrchestrator), preventing provider streaming errors; the registry logs registrations skipped for a missing tool contract. (PR #5903)

• 🐛 WP_Error Fatal Fixes in the memory layer and REST chat-memory paths (mine_agent_memory, retrieve_agent_memory, wake_up_context, request guard). (PR #5905)

• 📦 nvoos-content-graph v1.0.3 — wp.org resubmission with WPCS/review fixes and .wordpress-org screenshot assets. (PRs #5897, #5899, #5904)

✨ What's New at a Glance (v1.1.59)

• 🕷️ Media Worker Crawling & Crawl4AI Facade (v3.2.0). New /api/crawl/* endpoints — single-URL Markdown, batched crawling (sync or queued async), and link scans — with a two-tier extraction pipeline (static Readability → Turndown first, hardened Chromium fallback) behind the shared SSRF guard. A Crawl4AI-compatible facade lets the plugin's run_crawl4ai_job remote mode target the worker directly, with LLM-based structured extraction. Toolkit memory estimate now accounts for the worker sidecar. See docs/project/proposals/031-media-worker-crawl4ai-integration-plan.md. (PR #5892)

• 🔍 Research Tools Multi-Provider Hardening. semantic_content_search resolves embeddings through the shared provider abstraction (OpenAI, Gemini, Ollama, DigitalOcean) independent of the chat provider, falls back to keyword search, and skips mismatched-model vectors; new Gemini embedding provider. deep_research retries empty/truncated completions across the provider chain and never caches empty reports. Two new read-only base tools: list_terms and list_taxonomies. (PR #5893)

• 📚 Docs Hub Rebuild & Broken-Link Fixes. The plugin updater now fires wp_mcp_ai_plugin_updated after in-place updates; Docs Hub 0.4.1 auto-clears the cache and enqueues an async rebuild (plus an admin_init version-mismatch guard). Broken-link detection resolves links against the slug map, and suggestions are case-insensitive with clamped confidence. (PR #5894)

• 🔧 Tool Registration Fixes. Legacy-format tool classes are auto-wrapped, ~32 previously-orphaned base tools are now registered, and the new wp_mcp_ai_tools_init action gives side-loaders a late registration point. QA container memory limits raised. (PR #5895)

✨ What's New at a Glance (v1.1.58)

• 🔌 Composio Connect (Pro). New addons/pro/includes/composio/ subsystem bridges WordPress to the Composio MCP gateway: OAuth connection flow with state nonce, API client, trigger bridge, and signed webhook controller. Six new beta tools — composio_list_tools, composio_get_tool_schema, composio_list_connected_accounts, composio_create_connect_link, composio_execute_tool, composio_manage_triggers — plus remote-sites admin and metabox panels. See docs/composio-connect.md. (PR #5889)

• ⚙️ OOS Runtime Consolidation (Phases 0–5.8). The OOS engine now reaches parity with the legacy path — tool-surface and security-gate parity, an event-sourced session log, opt-in shadow mode (parallel parity runs serving the legacy result, zero user exposure) with wp mcp-ai oos parity CLI, canary routing per assistant, scoped tools with a compaction seam, Pro composition & child binding (wp mcp-ai composition), and single-path telemetry. See docs/project/proposals/029-oos-orchestration-runtime-consolidation-implementation-plan.md. (PR #5881)

• 🏷️ Standalone Plugins Renamed to Content Graph. plugins/nvoos-graphify → nvoos-content-graph, nvoos-graphify-ai → nvoos-content-graph-ai, nvoos-graphify-ai-platform → nvoos-content-graph-ai-platform (v1.0.2) with matching workflow and build-script renames. The addons/graphify/ knowledge-graph addon is unchanged. (PR #5890)

• 🐛 Security Center Refresh Fix. wp-api script now enqueued on the security tab — posture refresh, IP tests, snapshot restore, compliance export, and self-test no longer throw apiFetch is not defined. (PR #5887)

• 🔒 deepmerge-ts CVE-2026-40345 overridden in the media-worker and Pro packages. (PR #5888)

✨ What's New at a Glance (v1.1.57)

• 🔄 Plugin Updater Rework. New base-only update flow for wp.org base installs (check + in-place install via the GitHub base package; new "Base Version" panel in Settings → Advanced). Installation no longer uses Plugin_Upgrader — updates are copy-in-place with a backup snapshot and automatic rollback, so ZIP top-level folder-name mismatches can't break updates. Pro version now read from the plugin header (get_pro_installed_version()), fixing drift in the admin page and wp mcp-ai pro status. (PR #5871)

• 🤖 Hermes Chat — Async Submit/Poll. hermes_chat now submits via /api/chat/start and polls /api/chat/stream/status — runs keep executing server-side if the MCP client drops, timeout expiry returns still_running + stream_id, and the approval gate is answered with the configured HERMES_APPROVAL_MODE choice. Fixed answer extraction for live WebUI payloads. (PR #5872)

• 🐛 Service Status Provider Detection. AI-provider detection now resolves credentials through WP_MCP_AI_Credential_Resolver (plugin settings, WP 7.0 Connectors, env vars, PHP constants) instead of raw settings keys; new 135-line test suite. (PR #5874)

• 📚 Fleet Operator Agent Context. New addons/fleet-operator/.context/ tree (18 files) — addon conventions, Hermes ops, MCP integration, 6 operator roles, and task/memory templates for operator-agent sessions. (PR #5873)

✨ What's New at a Glance (v1.1.56)

• 🏢 Media Worker v3.0.0 — Multi-Tenant Shared Worker Mode (v2.4.0). SITE_TOKENS per-site isolation (fail-closed auth, namespaced files/queues/rate limits), per-site rate-limit overrides, SITE_TOKENS_PREVIOUS rotation. (PR #5866)

• 🔑 Phase 2 — Per-Site Provider Keys & Observability. SITE_PROVIDER_KEYS per-tenant AI provider credentials (with PROVIDER_KEYS_STRICT fail-closed mode), per-site usage counters, grouped temp TTLs, cluster-mode boot warnings, and a k6 load-test kit with split decision table. (PR #5868)

• 📐 Phase 3 — Scale Without Breaking Changes. Multisite per-blog worker tokens, plugin-side Media Worker Usage Reporter (opt-in daily cron), SITE_TOKEN_<SLUG> / SITE_PROVIDER_KEYS_<SLUG> env merges, opt-in Redis rate-limit store (RATE_LIMIT_REDIS=1), PROVIDER_KEYS_FILE hot-reload, manual Velocity deploy workflow. All opt-in — defaults unchanged.

• 🔄 Worker Routing Expansion. Document generation, OCR, video frames, health charts, email, QR/translate/PDF, and vectorization now route through the connected worker with local fallbacks; Pro settings lists worker-routed packages; new probe-wordpress.php connectivity probe.

• 🔐 Rotation, Canvas v3, Cloudways. Zero-downtime token rotation (WORKER_API_TOKEN_PREVIOUS), Canvas v3 napi prebuilds (compiler-free installs), Cloudways readiness hardening, live route bug fixes. (PRs #5863–#5865, #5867)

• 🤖 Hermes Tooling. WebUI MCP server (bin/hermes-mcp-server.js) to drive a Hermes agent from Zed (sessions, synchronous chat, skill sync), SSH MCP bridge, hardened env-file parser, and the Zed Hermes Console profile. (PR #5862 + follow-ups)

✨ What's New at a Glance (v1.1.55)

• 🔧 MCP Agent Compatibility & Reliability. JSON-RPC errors now return HTTP 200 so agent SDKs that drop non-2xx bodies relay tool errors instead of hanging. New legacy HTTP+SSE transport with credential-bound sessions for SSE-only clients. Tool rate limiter is settings-driven with credential-token exemption; GET/HEAD no longer consume the request quota; raw cred_* authorization headers accepted; async tool polling bounded at ~45s with inline kick of stuck jobs. (PR #5859)

• 🚀 Hermes Fleet Operator Addon. New addons/fleet-operator/ — scoped op_ operator credentials with audience binding, expiry, rate limits, and instant revocation so a supervisor agent (Hermes or any MCP/A2A host) can operate the site within an allowlist. Server-side tools/list scoping + tools/call enforcement. Ships admin page, WP-CLI commands, Hermes config generator, 3-skill nvoos pack, runbook, and tests. (PR #5858)

• 🔒 Media Worker v2.2.0 Security Hardening. Timing-safe X-Site-Token auth, SSRF guard, sandboxed Puppeteer launcher, rate limiting, Helmet headers, split health endpoints, request logging, graceful shutdown. WP_MEDIA_WORKER_TOKEN constant support in the plugin sidecar. Cloudways Velocity deployment guide + CI. (PR #5857)

• 🗄️ Database Connection Pooling Stance (Proposal 023). RabbitMQ-aware gating of Action Scheduler fallback and DB polling cron, atomic concurrency-guard slot tracking (mcp_ai_concurrency_slots table + daily cleanup), PDO persistent connections in Graphify, configurable queue worker batch size, and Site Health checks for MySQL pool, queue depth, and RabbitMQ health. (PR #5855)

• 📦 PostCSS 8.5.26 (GHSA-6g55-p6wh-862q) and media-worker subtree sync workflow (PR #5856).

✨ What's New at a Glance (v1.1.54)

• 🔒 PostCSS CVE-2026-69153. postcss minimum bumped to 8.5.23 in package.json overrides across 2 addons. Resolves a high-severity CVE in the CSS post-processor chain. (PR #5850)

• 🐛 MCP Async Tool Response Handling. tools/call endpoint now correctly handles async tool responses — the response handler was dropping results for tools that return promises or deferred execution. +111 lines. (PR #5845)

• 🔧 Plugin Updater Integrity Check v2. Follow-up hardening to the post-install integrity check (v1.1.52): fixed phantom bridge/ file false-positive from stat cache, cleared PHP clearstatcache() after each file check, and handled main-file rename edge case. (PR #5846)

• 🔑 API Key Merged-Settings Fix (Research Tools). deep_research, web_search, and 18 other research tools were reading API keys from the global option instead of the merged provider-specific settings. Now uses get_merged_credentials() so per-assistant and per-provider API key overrides take effect. 20 files, +37/-33 lines. (PR #5852)

• 🧠 Design Skills Audit & Enhancement. All 22 design-* skills comprehensively enhanced with detailed operational instructions, cross-references, and tool-calling patterns. 7 new pro-toolkit skills created: design-ai-assistant-admin, design-crm, design-project-management, design-communications, design-services, design-team-management, design-vault, and design-security-ops. Total: 29 enhanced/created skills, ~8,000 lines of agent-facing content. (PR #5847)

• 📋 OKF YAML Frontmatter Compliance. Added missing type: Skill frontmatter to all 22 design-* skills (PR #5844). Fixed YAML frontmatter spec compliance for 9 design skills — removed duplicate/invalid top-level keys that violated the OKF v0.2 schema. (PR #5849)

• 📚 README TOC Anchor Fixes. Fixed broken TOC anchor links caused by Unicode emoji rendering differences (VS16-based emojis like 🪲🐛 and U+26xx/U+27xx symbols like ⚠️⚡) across GitHub's markdown renderer. (PR #5853)

• 🧹 Stale Build Artifacts. Removed 30 stale v1.1.52 build artifacts from the repository. (PR #5851)

✨ What's New at a Glance (v1.1.53)

• 📊 Shared Analytics Service. New unified analytics subsystem (addons/pro/includes/analytics/, 1,489 lines) consumed by all Pro toolkits. 7 platform adapters: Meta (FB+IG), Twitter/X, LinkedIn, TikTok, WooCommerce, Google Analytics 4, and Cloudways monitoring. 5 immutable DTOs (Account, Post, Metric, TimeSeries, Report). Cross-platform normalization maps platform-specific metric names to a unified schema. Smart caching with stampede prevention. Token-bucket rate limiter per platform. Extensible adapter pattern — register new platforms via $service->register_adapter(). MCP server registration and Site Health integration. Design Stack skill: design-analytics-reporting. (PR #5836)

• 🛡️ Circuit Breaker & Execution Pipeline Hardening. Circuit breaker protection added to all remaining AI provider clients — now active on all 15 first-class providers with configurable failure thresholds. Concurrency guard, cost tracker, and backpressure signals wired into the execution pipeline. The agentic loop now auto-throttles tool calls when system load is high. Plugin updater base-only UI fix after complete upgrade. Silent build asset failure now surfaces a clear error. (PR #5839)

• 🧠 Agent Skills Sync & Discovery. sync-agent-skills now syncs project-owned design-* skills alongside wp-* skills — 22 new coding-time skills (design-analytics-reporting through design-video-creation) auto-discovered by the Zed editor. New mcp-ai-wpoos-plugin bundled skill. Base bundled skills increased from 45 to 67. (PR #5842)

• 🔧 SSE Stream & Load Guard Fixes. SSE backoff counter now resets between successful stream reconnections. Rate-limit counter properly cleaned up after stream termination. Load Guard fatal error on PHP 8.1+ fixed — no longer calls private count_active_jobs(). (PRs #5841, #5840)

• 📚 Documentation Catch-Up. CHANGELOG v1.1.53 entry documenting all missing features (Shared Analytics, circuit breaker, skills sync). README, CLAUDE.md, AGENTS.md, and 6 .context/ files updated with current counts, review dates, and architecture details.

✨ What's New at a Glance (v1.1.52)

• 📄 Paper Store Remote Site Support. All 8 Paper Store tools now accept an optional connection_id to proxy operations through Remote Site Manager. New REST API controller at mcp-ai/v1/paper-store (697 lines) enables remote CRUD. New WP_MCP_AI_Paper_Store_Remote trait handles dispatch. New list_mcp_tools discovery tool (234 lines) for AI agent self-discovery — filterable by toolkit and search term. (PR #5835)

• 🔌 Remote Connection CPT Auto-Discovery. Plugin-registered custom post types (including paper_store) now appear automatically in remote connection post type access controls. No more manual slug entry. resolve_post_type_access() mirrors discovery logic on form submission. Multi-site gateway plan (1,429 lines) for hub-and-spoke federation targeting v1.5.0+. (PR #5834)

• 🎨 Design System Tool Preset. New design-system preset with 72 tools across 13 categories: WordPress content, image generation/analysis/processing, web search/research, charts, document generation (14 tools), Paper Store (8), video generation/processing, social media, AI/memory, and utility. (PR #5837)

• ✅ Post-Install Integrity Check. Plugin updater now verifies 15 critical file paths after every update before reporting success. Prevents silent corruption on distributed filesystems (Cloudways). REST controller require_once calls now guarded with file_exists() checks. (PR #5833)

• 📚 Documentation Catch-Up. 3 new feature docs: Paper Store, Remote Sites & Connections, MCP Protocol Version Negotiation. Updated: Tool Presets System (Design System preset), Plugin Updater (integrity check), DOCUMENTATION_INDEX, QUICK_REFERENCE, README, CHANGELOG.

• 🐛 Fixes. Pro addon update visibility for wp.org base + standalone Pro (PR #5832). Docker chmod warning suppression (PR #5831).

✨ What's New at a Glance (v1.1.51)

• 🏗️ Orchestration & Harness Layer Gap Remediation. Comprehensive audit of AI orchestration and LLM harnessing against August 2026 industry standards (OWASP LLM Top 10, EU AI Act, NIST AI RMF, MCP/A2A protocols, OpenTelemetry GenAI). OWASP LLM Top 10 coverage: 20% → 60%. EU AI Act compliance: 17% → 67%. 6 new classes: Output Guardrail (OWASP LLM05 — validates LLM responses for sensitive info/unsafe content), Citation Verifier (OWASP LLM09 — cross-references claims against sources), Model Integrity Verifier (OWASP LLM03 — supply chain security with blocked model list), Semantic Cache (two-tier: exact + embedding-based prompt caching), Canary Deployment (progressive model rollout with auto-rollback). Workflow Engine V2 graduated from feature-flag to GA. Red Teaming scripts (shell + PHP) for OWASP-aligned adversarial probes. 3 new docs: Orchestration & Harness Reference, EU AI Act Compliance Mapping, OWASP LLM Top 10 coverage matrix. (PR #5830)

• 🔄 MCP Protocol Version Negotiation. Older clients (Zed, Claude Desktop, Cursor) rejected the hardcoded 2026-07-28 protocol version. Server now negotiates the highest version both client and server support, falling back to 2024-11-05 when the client provides no version information. 2 files, +123 lines. (PR #5829)

• 🔒 Security Bumps — Media Worker. multer 1.4.5-lts.1 → 2.2.0 (DoS vulns patched, 1.x deprecated). nodemailer 8.0.5 → 9.0.5 (GHSA-p6gq-j5cr-w38f raw bypass fixed). sharp 0.33.0 → 0.35.3 (libvips CVEs CVE-2026-33327 etc.). No code changes — APIs are unchanged across all three bumps. (PR #5828)

✨ What's New at a Glance (v1.1.50)

• ⚙️ Media Worker Sidecar. Docker-based Node.js sidecar with 11 Express route handlers: browser (Puppeteer), code (sandboxed), data (ETL), document (generation), email (rendering), image (manipulation), ocr (Tesseract), pdf, social (cross-platform), video (transcoding), and workflow (pipeline orchestration). Queue module with concurrent processing and retry logic. Pro integration via WP_MCP_AI_Media_Worker_Settings + Media_Worker_Client trait for tool-to-sidecar communication. Docker DNS-to-IP loopback resolution for container networking. Pro services updated (Fluent FFmpeg, Language Detection, MJML, Nodemailer, OCR, Prettier, Video Frame Extractor). (PRs #5822, #5823)

• 🐛 Site Health Redeclaration Fix. Fixed PHP fatal error from function redeclaration in ensure_site_health_dependencies() — now guarded with function_exists(). (PR #5824)

• 🔒 NPM Security Fixes. nanoid, js-yaml, and dompurify vulnerabilities resolved across 11+ addon package-lock.json files. (PR #5821)

• 🤖 BMAD Agent Editing Conventions. Developer and QA agent YAMLs updated with code editing conventions for consistent agent behavior across sessions. (PR #5825)

• 🧹 WPCS Formatting Cleanup. Repository-wide PHPCS fixes: tabs, docblock formatting, and code style corrections across all PHP files. No functional changes.

✨ What's New at a Glance (v1.1.49)

• 🔧 Gemini Model Resolution Fix. Gemini client now uses the correct settings key instead of the deprecated gemini-pro fallback. All model lookups routed through the canonical provider settings path. (PR #5817)

• 🔄 Update Reactivation & Release ZIP Cleanup. Plugin updater now correctly reactivates after update. Nonce-scoped update actions hardened. Dev files (.agents/, .context/, .github/, IDE configs, test fixtures) excluded from release ZIPs via .distignore and .gitattributes. (PR #5816)

✨ What's New at a Glance (v1.1.48)

• 🛒 Shopify Sync Toolkit — 7 Bug Fixes. Fatal error (graphql_query() → graphql()) on Admin API connection test. Dead code in list_by_status replaced with direct SELECT COUNT(*) query. Always-zero analytics count now uses actual API response edges instead of empty array. Duplicate get_option() call and redundant $sync_mode assignment removed. Infinite loop risk in truncate() eliminated — paginated while-loop replaced with direct DELETE FROM {table}. Double DB query per upsert halved via by-ref &$operation output parameter. Missing orderby/order schema parameters added with sanitization passthrough. (PR #5815)

• 🔒 PHPCS CVE-2026-67434. squizlabs/php_codesniffer bumped 3.13.4 → 3.13.6 across 3 composer.lock files. Resolves arbitrary code execution via crafted ruleset XML. Dev dependency only — no functional impact.

• 📚 Default Skill Catalogues. Brave Search Skills and all 22 WordPress Agent Skills added to the default skill catalogue for new assistants out of the box.

• 🛠️ NVoOS Graphify Standalone Plugins. 3 new standalone plugin ZIPs in plugins/ directory: Graphify v1.0.1, Graphify AI Platform v1.0.0, and Graphify AI v1.0.0.

✨ What's New at a Glance (v1.1.47)

• 🛢️ MySQL Connection Exhaustion Fix (Cloudways). Cron system overhaul with unified concurrency limits, per-process memory caps, and staggered offset scheduling to prevent connection-pool saturation. Activation bootstrap now uses connection throttling. Service status registry hardened. Pro status page JS, AJAX handler, and dashboard improvements. 7 files, +438/-89 lines. (PR #5809)

• 🔄 Update Checker Cache Bust. Manual "Check for Updates" now clears cached release data and fetches live from GitHub Releases API. (PR #5810)

• 🔒 Mermaid Security Fixes. npm audit fix in canvas-toolkit: mermaid 11.15.0 → 11.16.1 resolves 5 CVEs (prototype pollution, CSS injection, DoS). (PR #5812)

✨ What's New at a Glance (v1.1.46)

• 💾 Comprehensive Backup & Restore (Proposal 020). 11 modular export providers (8 base + 3 Pro) with JSON-based export/import. Admin UI with provider checkboxes and progress feedback. Covers settings, assistants, CPTs, custom tables, federation, addon options, toolkit options, JetEngine CCTs, license keys, and remote sites. +3,660 lines. (PRs #5805, #5807)

• 🔄 GitHub-Based Plugin Updater. New WP_MCP_AI_Plugin_Updater class fetches releases from GitHub API, installs via WordPress Plugin_Upgrader. Base-to-complete upgrade path in Settings → Advanced. Pro addon update support with rollback. Cache bust on manual check. +772 lines. (PRs #5800–#5804)

• 🔌 Abilities API Selective Adoption (Proposal 019). New includes/abilities/ framework — Registrar, Bridge, Category Registrar, and Security Bridge for machine-readable plugin operations with JSON Schema contracts. Tool-ability interface for MCP/AI agent discovery. 5 test files. (PR #5799)

• 📚 6 New Default Skill Catalogues. Brave Search Skills, WordPress Agent Skills, Cloudflare Agent Skills, Google Workspace CLI, OpenAI Agent Skills, and Google Agent Skills added to default skill catalogue for new assistants out of the box.

• 🐛 Status Page Fixes. Fatal error in Pro status REST endpoint resolved. JS errors and i18n text domain consistency fixed on the status dashboard. (PR #5806)

• 📦 Knowledge Base Auto-Build. CI auto-builds knowledge-base.zip when playbook files change. (PR #5797)

• 🧹 PHPCS Cleanup. Parse error fixes from canonical envelope conversion, text domain mismatch corrections, and WPCS formatting across 100+ tool files. (PRs #5798, #5804)

✨ What's New at a Glance (v1.1.43)

• 🔒 Security Hardening v1.1.43 (16 files). SSRF protection across 7 provider connection handlers + A2A agent URL. SQL table-name validation against injection. A2A per-assistant agent cards now auth-gated. Chat SPA /config and /manifest endpoints require authentication (no longer public). Missing args schemas on 7 REST endpoints. Guest rate-limiting now IP-based (DoS vector closed). 3 tool capability mismatches corrected. Centralized defense-in-depth capability gate in Tool Registry. PHP object injection prevention via safe_unserialize helper.

• 🆙 MCP Protocol Upgrade (2026-07-28). Stateless core — sessions retired, server/discover replaces initialize, _meta per-request capabilities, Mcp-Method/Mcp-Name headers on Streamable HTTP, TTL/cache-scope on tools/list. Legacy client shim for backward compatibility. 10 files changed.

• 📚 OKF v0.2 Trust-Signal Support. Indentation-aware recursive descent parser with inline YAML mappings, nested object lists, and flow sequences. New trust-tier derivation (unverified/machine-confirmed/human-reviewed), staleness checks, search() extension with trust filters. New okf_validate_attestation tool. All 6 existing OKF tools surface trust signals. 15 files, 27 standalone smoke tests.

• 🏗️ Pro Module Registry. wp_mcp_ai_pro_init() decomposed from ~625-line monolithic function into a PSR-4 module registry with per-module loading, dependency ordering, and graceful degradation. Fixes broken require paths in module loader.

• 🎯 Ideal Customer Profile (ICP) System (Pro CRM Phase G). 7-dimension scoring engine (0–100) with fit+intent separation, behavioral decay, and negative scoring. 2 new MCP tools: compute_icp_score, manage_icp_profile. Admin UI with profile list table and 7-tab editor.

• 🔧 Profession & Playbook Sync Fixes (7 PRs). Playbook bulk sync silent failures now report errors. Force-regenerated playbook files no longer deleted. Undefined array key in playbook stats fixed. Duplicate profession slugs removed from knowledge base (312→311 entries). Title fallback prevents duplicate posts on Update. Consumed-post collision fix ensures all unique KB entries create distinct posts. Test Model assistant dropdown restored.

• 🧹 Hexagonal Architecture Purity. PlatformFlushInterface contract extracts wp_ob_end_flush_all from framework-agnostic lib/core/ SSE handler into a WordPress adapter — zero WordPress references remain in nvoos/core.

• 🛡️ Phase 3 Operational Security. Audit logger REST route hook corrected (init → rest_api_init). Security posture signals hardened.

• 📦 Dependency Updates. WPCS bumped to 3.4.1 (CVE-2026-45293). Dev tooling: Phpactor LSP replaces Intelephense.

✨ What's New at a Glance (v1.1.42)

• 🛡️ Security Infrastructure Hardening. 7 new security infrastructure classes (Request Guard, Security Posture with 21 signals A-F grading, Destructive Ops Gate, URL Guard, Concurrency Guard, Cost Tracker, API Key Store). Site Health integration. Production hardening guide. CORS, rate limiting, error verbosity, and body size enforcement with admin dashboard posture signals.

• 🏗️ Framework-Agnostic Core (lib/core/). nvoos/core package — Hexagonal Architecture with 32 domain contracts + 21 WordPress adapters. ChatOrchestrator with RateLimiter + SemanticCompressor. 109 tools migrated to framework-agnostic format. ProviderRouter with 12-provider routing. 5 chat parity gaps closed.

• 📡 Status Page & Incident Communication (Pro). Maintenance window system with frontend banner, countdown timer, and multi-channel notifications. Incident workflow with phase state machine, phase-aware dispatcher, and lesson bridge. 4 new AI tools for service status and incident management.

• 🤖 Agent Skills & BMAD Agents. 21 coding-time agent skills for Zed editor covering WordPress plugin development patterns. 6 BMAD workflow agent YAML definitions with team composition config. Full .context/ subsystem context files with 8 topics + 5 templates.

• 🎵 Algorave Addon. New live coding and algorithmic music generation addon with 9 tools, pattern/session CPTs, and REST API.

• 🔒 Security Hardening (12 fixes). OAuth token lifetime controls, DICOM PHI auto-redaction, asset version stripping, exception guard, auth brute-force protection, SSE CORS filter, webhook secret indicator. 13 new security unit tests.

✨ What's New at a Glance (v1.1.41)

• 📚 OKF Integration (Open Knowledge Format v0.1). Google vendor-neutral knowledge format engine with 6 MCP tools for curated, deterministic knowledge management. All 41 bundled skills are now OKF v0.1-conformant.

• 🔒 Security Compliance Fixes (11 HIGH/P0). HMAC-signed policy tokens for professional selector, health endpoint auth-gating, ZIP path traversal prevention, CSRF nonces on sync endpoints, SRI integrity hashes for all 6 CDN libraries, Google Chat OIDC hardening.

• 🔧 Playbook Sync Fixes. Duplicate AJAX handler conflict resolved. Silent sync failures now report errors to admin UI. CPT class loading guards prevent fatal errors during bulk reseed.

• 🔑 Model Provider Credential Resolution. Model picker now resolves API keys from all 4 sources — settings, credentials option, environment variables, and PHP constants.

• 🛡️ Dependency Security. adm-zip, axios, and brace-expansion bumped to resolve 18 Dependabot alerts across 5 package.json files. npm audit: 0 vulnerabilities.

✨ What's New at a Glance (v1.1.40)

• 📝 Content Format Awareness. New WP_MCP_AI_Content_Format_Helper detects and preserves Markdown, HTML, and plain text formats across post-modifying and analysis tools — AI-generated content retains its intended structure through the full create/update pipeline.

• 🔗 Research → Paper Store → WordPress Draft Pipeline. All research tools now support save_to_paper_store for staging results in the flat-file Paper Store. New create_post_from_research Pro tool bridges staged research to WordPress drafts. Human-in-the-loop review before publishing.

• 🎬 Demo Video Pipeline Complete (Phases 0–5). Scripted scene recording with AI voiceover narration. GitHub Actions CI workflow automates video assembly. 14 narration scripts and video catalog.

• 🔐 Settings Credential Split. Sensitive API keys moved from wp_mcp_ai_settings (autoload) to separate non-autoload wp_mcp_ai_credentials option with transparent merge. One-time migration. wp_suspend_cache_addition and dual cache clearing for defense-in-depth.

• 🤖 Kimi & DeepSeek Client Parity. Both providers now match all first-class providers: streaming, tool use, token tracking, and error handling. Plus DeepSeek and 9 missing providers added to all research tools.

• 📊 Model Catalog Update (July 2026). 24 files updated across base + pro. Default bumps: Gemini gemini-2.5-flash → gemini-3.5-flash, NVIDIA meta/llama-3.1-8b → nvidia/nemotron-3-nano-30b-a3b, Gemini Live gemini-2.5-flash-live → gemini-3.1-flash-live-preview.

• 🏗️ OOS Engine: SchemaStoreInterface + Tests. New domain contract, PostTypeSchema/TaxonomySchema entities, GetPostTypeSchemaTool, WordPress adapter. 45 new tests.

• ⚡ SSE HTTP/2 Fixes. ob_clean() replaces ob_flush() to prevent protocol errors. 524 timeout resolved in Pro SPA v2.

• 🔍 Vector Store Sync — No Polling. Status now checked only on assistant change and page load, reducing API load.

• 🔧 Settings Import/Export Batch (4 fixes). Credential merge, save key wipe, subtab sanitization, export consistency — all resolved.

• 🛡️ Validated Tool Slug Allowlist. Tool slug matching fixed for validated variants.

✨ What's New at a Glance (v1.1.39)

• 🤖 Page Agent Addon v0.1.0. New addon (addons/page-agent/) — AI-powered browser page control copilot powered by Alibaba Page Agent (MIT). Give any WordPress page its own AI agent that can click, type, and navigate via natural language, running entirely client-side with no headless browser required. Includes shortcode, Elementor widget, REST endpoints, and MCP tool bridge.

• 💬 Pro SPA v2 — Major Parity Update. Voice pipeline, tasks drawer, workflow tracker, and file attachment upload to WordPress Media Library. Tool Shortcuts and Slash Commands drawers. Mobile hamburger sidebar toggle. Speech/audio button fixes using correct REST endpoint. Conversation title improvements and turn count display fix. Button and token NaN display fixes. Model sync and auth bypass fixes. Assistant preloading in runtime config.

• 🎨 Pro SPA v2 — UI Polish. Autoscroll fixes (submit, streaming start, user-at-bottom guard, scrollTop vs scrollIntoView). Viewport height fixes via CSS height chain instead of viewport calc. overflow:hidden on height-chain ancestors. filemtime cache-busting across all SPA addons. Lint errors resolved.

• 👤 Per-User Chat Memory Preferences. Users can now toggle chat memory on/off from their WordPress user profile. Individual control over AI memory retention without affecting site-wide defaults.

• 📝 create_post / save_post Tool Enhancements. Markdown-to-HTML conversion via new WP_MCP_AI_Tool_Markdown_Converter trait. Smart taxonomy suggestions auto-detect relevant categories and tags. Block content corruption fixed for non-post post types.

• 🔄 Workflow Blueprint & Schedule Improvements. Existing-content awareness in Content Publisher and Keyword Pipeline blueprints. Blog schedule presets now check for duplicate content before publishing. Readable response generation for workflow schedule result delivery.

• ♿ SPA Accessibility. Annotation pills made clickable with meaningful screen-reader labels.

• 🛡️ Security. OWASP ZAP DAST medium findings triaged as false positives.

✨ What's New at a Glance (v1.1.37)

• 🏭 EZuite Inventory Sync Pro Toolkit. ERP-integrated inventory sync bridging EZuite ERP with WooCommerce/WordPress. Pull products, query inventory, create/update items, manage orders, and configure API credentials — all via AI tools. Admin UI with connection selector, field mapping, and sync direction controls. CLI sync commands for batch operations.

• 🔄 Ralph Loop CCT Migration & Orchestration. Circuit breaker pattern with configurable failure thresholds. Execution logger with step-by-step tracking. CCT migration tools for safe cross-environment JetEngine data operations. Orchestration tools for multi-step workflows.

• 📅 JetBooking/JetAppointment Integration (8 tools). Adapter layer for Crocoblock booking and appointment plugins. 8 new tools + 4 enhanced calendar tools with booking/appointment awareness.

• 🧠 Moonshot AI (Kimi) & Z.AI (GLM) Provider Parity. Both providers upgraded to full DeepSeek-level chat client capabilities — streaming, tool use, token tracking. ZAI client + baseten service in DI container. Provider count now 15 first-class.

• 📋 Unified Sync Log Manager. Per-item audit trail across EZuite, FlowHub, and Shopify sync toolkits. Sync history with timestamps, status, and error tracking. Status dashboards on admin pages.

• 🎛️ Tool Presets Auto-Select & Chips Bar. Selected tools display as clickable chips with +N overflow toggle. Tool payload cap raised from 50 to 100 tools per assistant.

• 🌐 HTTrack Cache & Place-to-Service Bridge. HTML mirror import now supports HTTrack cache directories. Auto-creates bookable services during batch place import. URL backfill for mirrors without hts-cache.

• 🔌 FlowHub Per-Connection Overview. Remote Sites connection selector on FlowHub config tab. Per-connection sync controls. Proxy support via http_api_curl hook.

• 🔍 Web Search 429 Retry. Exponential backoff for rate-limited search requests.

• 🐛 45+ Bug Fixes. CCT module API mismatches across EZuite/FlowHub/Shopify (canonical Module.instance/Factory/ItemHandler). EZuite sync (missing return, field mapping, connections, CCT registration). FlowHub sync (silent failure, proxy persistence, auth headers, null cct, dry-run). Shopify sync (Catalog API guard, CCT registration lifecycle). Duplicate column errors in ensure_columns. SQLite meta cache explosion. masterminds/html5 case collision. Base-version guard blocking toolkits. Tool registry fatal guards. HTTrack import (URL resolution, hex filenames, subdirectory content, mirror detection). Necessity Gate request context crash. Auto-select compute timeout. Place-to-Service bridge collision.

• 📖 Documentation. Abilities registration plan (~1,000 tools as WordPress Abilities). Laravel-scale deployment architecture proposal. WP.org submission prep. Agent context sync.

✨ What's New at a Glance (v1.1.35)

• 🏪 FlowHub Inventory Sync Pro Toolkit. 6-tool cannabis dispensary management: products, inventory, locations, sync, analytics, alerts. P1 proxy support via http_api_curl hook. P2 CCT auto-registration in JetEngine custom content type tables. Auth, decryption, location_id, and null-guard fixes (PRs #5500, #5501, #5502, #5503, #5507, #5510). Admin UI toggle on Features tab.

• 🛒 Shopify Sync Pro Toolkit. 5-tool bi-directional e-commerce sync: products, orders, inventory, analytics, settings. Connection resolver trait. Dashboard widget with sync status and recent activity. Tool reference docs at docs/tools/shopify-sync-toolkit.md (PR #5502).

• 🛡️ Necessity Gate Layer J. Pre-execution safety layer that scores tool calls by irreversibility risk. Write operations (create/update/delete) assigned risk scores by resource type. Safety profile trait with clean autoload. Request context crash fix.

• 🎙️ Local Voice Embedded STT. Three pluggable browser-side speech-to-text backends: Web Speech API, Whisper.cpp (WASM), Vosk (WASM). Offline-first — no server dependency. Auto-detection based on browser capabilities (PR #5498).

• 🌐 Remote Site Administrator Blueprint. 22-tool assistant blueprint in Site Creator toolkit for full remote/local WP/WooCommerce management with JetEngine, JetFormBuilder, and REST API control. Auto-discovered by Unified Blueprints page.

• 📦 Places & Calendar Bulk Import. Batch import tools for Places and Calendar Booking toolkits — import multiple records in a single call (PR #5509).

• 💻 CLI site-import Subcommand. Multi-phase HTML mirror import for migrating static sites into WordPress — page structure extraction, content mapping, media sideloading.

• 🎤 Voice Realtime Auto-Detect. WebRTC/WebSocket auto-selection, duplicate message prevention, VAD threshold improvements (PR #5508).

• 🔧 Remote Connections Fixes. WordPress case handling normalized. FlowHub and Printful credential storage fixed (at-rest encryption). Printful connection type added (PR #5499).

• 🐛 7 Bug Fixes. Token-scoped assistant resolution (PR #5497), user_id empty fallback (PR #5495), credential token mapping (PR #5493), post type name lengths (PR #5484), OpenAI image deprecation cleanup (PRs #5489-#5491).

• 📋 Documentation. GPT-Realtime-2 upgrade proposal + 1,166-line implementation plan. FastAPI porting implementation plan (PR #5467).

• 🧹 Housekeeping. Stale build artifacts and toolkit-addons directory removed.

• 🔌 WP 7.0 Connectors Credential Integration. Credential_Resolver now integrated into all 17 AI client get_api_key() methods. Fallback chain: WP 7.0 Connectors → plugin settings → env vars → PHP constants. Credential source badges and WP 7.0 Connectors hints rendered in admin settings UI. All 13 provider API key field descriptions updated. Provider diagnostics show key source column. Settings health check counts credentials via resolver.

• 🧩 nvoos-graphify v1.0.0. Standalone nvoos-graphify plugin released at v1.0.0 (Plugin Check compliant). nvoos-graphify-ai released at v1.0.0-dev. Fixed critical ->prepare() spread-operator bug in Db::listNodes(). Renamed vector column → embedding_vector to avoid MariaDB/MySQL reserved-word conflict. Fixed snake_case→camelCase method calls across tools, controllers, and cross-plugin integrations.

• 🛡️ Security Dependencies. guzzlehttp/guzzle 7.10.0 → 7.12.1 (CVE-2026-55568, CVE-2026-55767). guzzlehttp/psr7 2.11.0 → 2.12.1 (CVE-2026-55766). guzzlehttp/promises 2.3.0 → 2.5.0. undici npm override tightened >=7.28.0 → >=8.5.0.

• 🔒 npm Security. 29 alerts resolved across 14 packages: undici (TLS bypass CVE-2026-9697, cache info CVE-2026-9678), http-proxy-middleware (CRLF injection CVE-2026-55603), nodemailer (GHSA-p6gq-j5cr-w38f), webpack-dev-server (HMR CVE-2026-9595), dompurify (GHSA-cmwh-pvxp-8882). Fixed critical duplicate overrides key in root package.json.

• 🪲 Bug Fixes. Fixed missing -pro- in WP All Import/Export require_once paths causing fatal errors when those Pro tools load. Fixed fragile @file_get_contents() warning suppression in tool status label loader — replaced with explicit set_error_handler('__return_true') to prevent leaked warnings from corrupting MCP JSON-RPC HTTP responses.

• 📦 Dependencies. 15 Dependabot bumps across Composer, npm (stripe, zod, p-queue, csv-parse, react-query, puppeteer, vitest, wrangler, eslint-plugin, workers-types, types/node), and GitHub Actions (codecov 4→7, action-gh-release 2→3).

• 🧹 Housekeeping. Stale 1.1.31 and 1.1.32 build zips removed. SPA addon ZIPs rebuilt with updated security overrides.

✨ What's New at a Glance (v1.1.32)

• 📝 Content Format Templates & Featured Images. Content Format Template CPT with user-editable blog templates. Content Template Engine generates Anthropic-optimised AI prompts. Featured Image Service with 3-provider fallback (DALL-E → Gemini → Cloudflare) and 5 image styles. Provider image settings now respected instead of hardcoded defaults.

• 📬 Result Delivery Pipeline. 1,056-line service routes schedule results to 8 channels (email, Slack, Discord, Telegram, SMS, Paper Store, WordPress post, webhook). Both success and failure paths now deliver — previously only failures were surfaced.

• 📄 ECA Document Generation. ECA Consolidate & Add page with document generation tools.

• 🪲 Duplicate Posts Fixed. WordPress delivery channel removed from weekly_blog_post_writer and weekly_blog_topic_research presets. Systemic guard skips WordPress delivery when AI tool calls already include create_post or save_post.

• ⏱️ 6 Provider Clients Timeout Fix. DeepSeek, Baseten, DigitalOcean, OpenRouter, Kimi, and Cloudflare clients now respect the global request_timeout setting instead of hardcoding 60 seconds.

• 🔧 Schedule Trigger Stability. Trigger crash on rest_do_request() fixed with try/catch + pre-flight REST check. Result delivery sanitizer preserves channels wrapper so edit modal shows saved values. AJAX handler captures PHP warnings that corrupt JSON.

• 📋 Paper Store Delete Fix. Hidden inputs added to delete confirmation form so handle_delete_record() receives required POST fields.

• 🧪 npm CI & Jest Resilience. Babel ESM override pinned for Node 18/20 compatibility. Jest config probes for setup files at load time with graceful fallback. npm ci lockfile sync for 5 addons.

• 📦 Dependencies. 14 safe Dependabot bumps. 8 npm audit CVEs (nodemailer, tar, tar-fs). phpspreadsheet 5.7.0 → 5.8.0.

✨ What's New at a Glance (v1.1.31)

• 🎬 Media Command Center. Top-level NV Media admin menu with command center for media templates, presets, and blueprints.

• 💬 Pro SPA v2 — Rich Rendering & Scoping. Rich markdown rendering, per-assistant scoping, agent selector dropdown. Conversations primary, threads read-only. Version 2.0.1.

• 🧰 34 Workflow Preset Tools. All missing workflow presets implemented across 10 toolkits. Full PHPCS compliance and test suite for 36 tools.

• 🔒 npm Audit CVEs Resolved. 12 CVEs fixed: vite, launch-editor, markdown-it, ws, js-yaml, form-data, hono, dompurify, babel, opentelemetry, joi.

• 🎨 Gemini 3.1 Flash Image. Default Gemini image model upgraded to gemini-3.1-flash-image.

• 🖼 Media Toolkit Blueprints & Presets. Blueprints and scheduler presets synced to Data Management page.

• ⚡ OpenAI/DeepSeek stream_options. Streaming usage payloads now include proper stream_options for accurate usage tracking.

• 💰 Agentic-Loop Cost Tracking. Tool result costs computed from tokens when not explicitly provided. Missing provider pricing added.

• 🛡️ Vite CVEs. vite ^8.0.16 for CVE-2026-53571 (server.fs.deny bypass) and CVE-2026-53632 (NTLMv2 disclosure).

• 🧹 1,658 PHPCS Lint Fixes. Across base + pro addon (44 toolkits).

• 🔧 CI Disk Space. Free disk space step added to all build workflows.

✨ What's New at a Glance (v1.1.30)

• 💬 Chat SPA Phase 8 — Message Actions. Edit, delete, regenerate, copy, and content enrichment cards on every assistant message. Conversations sidebar with assistant scoping. Auto-create thread on first message.

• 📊 PM Toolkit Enhancement Phase A–D. Shared engine powering Command Center dashboard with real-time task visibility. Work Ingestion panel. 28 new AI tools for task management, resource allocation, timeline generation, and reporting.

• 🏢 CRM Toolkit — Duplicates & Hygiene. Duplicate detection engine with safe one-click merge. Email hygiene module: domain reputation classification, exclusion/priority lists, auto-pruning. Top Customers/Clients analytics. LinkedIn & Upwork external sourcing.

• 🥧 DietPi Pro Toolkit Phases 0–3. 19+ tools for server management: system info, package/service control, backup/restore, storage, provisioning, SSH proxy. Registered as MCP server.

• 🛡️ Layer I Guardrails — Jailbreak Prevention. Stay-on-target guardrails blocking prompt-injection and role-change attacks. Configurable per-assistant via LLM Harness metabox.

• 🪟 Context Window Management. Pre-flight validation across all 13 AI providers. tiktoken integration, estimator metabox, token-budget tool capping, chat parity drift detection.

• 🌉 WP 7.0 Connectors API Bridge. Forward-compatible provider credential bridge. Classes load unconditionally on WP < 7.0.

• 🧠 Chat Transcript & Agent Memory Retention. Configurable TTL-based transcript cleanup (base). Agent memory lifecycle management with pruning and retention windows (Pro).

• ⚡ Pro Toolkit Optimizations Phase 1–3. Performance optimization across 6 Pro toolkits with autoload control, caching, and lazy loading.

• 🔌 OAuth & API Disconnect Buttons. One-click disconnect for OAuth and API connections with automatic token clearing.

• 💬 LibreChat Addon. New standalone addon with code interpreter, speech-to-text/text-to-speech, and web search reranker.

• 🪲 30+ Bug Fixes. SPA reliability sweep (thread endpoints, router context, event serialization). Agentic loop tool result persistence. CPT slug limits. Security CVEs (guzzlehttp/psr7, shell-quote, esbuild).

✨ What's New at a Glance (v1.1.29)

• 🔧 Chat Bubble Assistant Dropdown. Settings UX fixed: chat_bubble_assistant_id changed from manual ID input to proper assistant <select> dropdown.

• 🪟 Context-Window Pre-Flight Validation. Added to all AI provider clients with shared validate_context_window() helper, tiktoken integration, and estimator metabox.

• ⚡ OpenAI SSE Streaming Fix. stream_options payload flag corrected so OpenAI real-time SSE streaming triggers properly.

• 📅 Schedule Preset Data Mismatches. Fixed presets losing configuration data after save. Improved error logging for preset operations.

• 🧹 Playbook Orphan Cleanup & Batching. Fixed orphan accumulation and sync timeouts with configurable chunk size batching.

• 🔄 Stale Provider Validation Lists. Provider checks moved to dynamic discovery instead of hardcoded arrays, fixing DeepSeek rejection.

• 📊 CRM Activity Titles, Due Dates & Block API v3. Activity post titles, recurring due dates, and admin blocks migrated to WordPress Block API v3.

• 🎯 OpenAI-Compatible Client — DeepSeek Parity. OpenRouter, Baseten, Kimi, DigitalOcean, NVIDIA NIM clients upgraded to DeepSeek parity.

• 🎤 Voice & Embedded LLM Missing Assets. Missing .min.js files added for voice recording/transcription and embedded LLM worker scripts.

• 🪲 25+ Additional Fixes. Chat config messagesEndpoint, debug console, OOS bridge fatal error, embedding service WP_Error, SSE header warnings, memory cookie nonce, graphify content leak, model limits sync, shell-quote CVE-2026-9277, and more.

• 🧪 Tests. Chat transcript REST controller tests improved from ~4% to 87% pass rate.

✨ What's New at a Glance (v1.1.28)

• 🏢 CRM Phase C Complete. IMAP email polling, Twilio SMS webhook, Meta WhatsApp webhook, and Gmail OAuth bridge for multichannel inbound ingestion — all triaged through CRM Classifier and routed to the Workflow Command Center.

• 👥 Customer CPT + Customer 360. 5 CRUD tools for mcp_ai_customer CPT. Customer Research & Add page with Customer 360 dashboard. Lead-to-customer conversion with deal promotion.

• 🎫 Support Ticket Module — 10 AI Tools + SLA. Full ticket lifecycle: create, get, update, list, classify, escalate, resolve, reopen, merge, SLA report. Ticket automation, SLA breach detection via cron, email notifications, and optional Zendesk sync.

• 🔍 TF-IDF + BM25 Relevance Search. Dual-algorithm relevance ranking across CRM, healthcare, and base content search tools. Shared traits in both Base and Pro.

• 🧠 Transformer-Inspired Attention Routing. QKV multi-head attention (5 heads: semantic, capability, recency, dependency, risk) for semantic tool selection. Sliding-window conversation compressor. Persistent tool embedding store. RRF fusion with harness scoring.

• 🔌 Funiq Bridge Addon. Payload-to-WordPress bridge with React admin SPA, REST controllers, transformers, post types, and taxonomies.

• 🕸️ NVOOS Graphify Ecosystem. Three standalone plugins: nvoos-graphify (visual knowledge graph, 14 tools), nvoos-graphify-ai (13 providers, streaming chat, RAG, embeddings), nvoos-graphify-ai-platform (Agents, A2A, ACP, Blueprints, Federation, Harness, Measurement, Professions, Skills, Slash Commands). Framework-agnostic lib/core and lib/wordpress-adapter packages.

• 🏗️ NV Platform AI Addon. Top-level admin dashboard + CPTs (Project, Resource, Template).

• 🎬 Automated Demo Video Pipeline (Phases 1–3). Scripted scene recording, AI voiceover generation, automated video assembly.

• 🛡️ CRM Lead/Deal Enhancements. Enriched lead/company tables, dedicated Leads tab, data completeness KPI. Lead CPT admin expanded with contact details and remote channel link.

• 📋 Documentation & Unix Theory. Folder READMEs for new CRM subdirectories (customers/, inbound/, support/). Compliance check errors in includes/data/ and addons/pro/includes/traits/ READMEs resolved. CRM enhancement plan updated.

✨ What's New at a Glance (v1.1.27)

• ⚡ Real-Time SSE Streaming. Real-time streaming enabled for OpenAI, DeepSeek, and all OpenAI-compatible providers. "Disable Native Streaming" control in Settings.

• 🧰 35 New OOS Core Tools. Data tools (GetPostTaxonomies, CountPosts, GetPostMeta, TruncateText, MergeArrays), format tools (FormatDate, TimeAgo, ParseCsv, MathEval, ColorConvert), infrastructure (EventDispatcher, Queue), and cache management tools.

• 📸 Extended Cognition Vision Recognition. Visual product/brand recognition with camera viewfinder UI, detection overlays, and consent gate.

• 🔧 JetFormBuilder Submission Tools — 8 Fixes. Empty results for non-admin users, form discovery pipeline, PHPCS warnings, REST route matching, form-type auto-detection, and plugin detection all fixed.

• 🎯 Graphify Tools Capability Compliance. Missing trait and explicit get_required_capability() added to all Graphify tools.

• 🤖 DeepSeek Agentic Tool Handling. Tool message filtering and payload normalisation for agentic multi-turn workflows.

• 📝 Docs Fixes. Broken links after Unix-theory reorganization resolved.

• 💰 June 2026 Model Pricing. All 13 provider pricing updated.

• 📋 Plugin Restructuring Proposals v3.0. Graphify-centric architecture spec and roadmap.

• 🛡️ Pro Toolkits Security Audit. 9 HIGH-severity security findings fixed.

• 📋 Reviewer Onboarding Docs. Complete reviewer documentation suite (docs/project/FOR_REVIEWERS.md).

• 🐳 Docker Dev Environments. WordPress, Laravel, and Craft CMS Docker environments all fixed.

• 🧪 Test Infrastructure. 95% of PHPUnit failures resolved across base, pro, and addon test suites.

• 🔧 Infrastructure Fixes. TCPDF autoloader fix, Pro vendor files committed, puppeteer detection path fix, shallow clone recommendation.

✨ What's New at a Glance (v1.1.25)

• 🧩 Unified Blueprint System. 55 pre-built AI assistant blueprints across 25 toolkits.

• ☁️ Cloudways Pro Toolkit. 60 AI tools for server and application management via Cloudways API v2.

• 🏢 CRM Toolkit Phases A–E Complete. 70+ tools: lead management, multi-channel triage, sequences, command center, compliance.

• 💬 Chat UI Enhancements. 7 features: profile card, stop generation, feedback, code copy, dark mode, prompts, search.

• 📂 Unix-Theory Tool Reorganisation Phase 4–5 Complete.

• 🎛 Pro Toolkit MCP Server Settings Pages. Phases A–C.

• 🏥 Aerlinn + Healthcare Blueprints.

• 🔧 Build Infrastructure Hardening.

✨ What's New at a Glance (v1.1.24)

• 🧹 Bug-Fix & Stabilisation Sweep. Paper Store Pro interface load order fix (deferred to wp_mcp_ai_bootstrapped hook). Chat SPA duplicate-message and SSE protocol-mismatch fixes. Markdown rendering enabled in Chat SPA responses.

• 🛡️ Skill Manager Canonical Envelope — Unix Theory P0/P1 Refinement. Skill manager now returns WP_Error on failure instead of the legacy array('success' => false, ...) pattern. Skills sync endpoint added for idempotent import/export. YAML frontmatter parsing hardened against colons in description fields.

• 📝 Assistant Tool Presets Coverage. 24 missing tools added to assistant creation presets. Out-of-date tests fixed.

• 🔒 CVE Patches. tmp bumped to >=0.2.6 and symfony/cache to ^6.4.40 to resolve upstream CVEs. Composer vendor state committed.

• 🗄️ Paper Store Admin CRUD. Full CRUD admin UI for Paper Store collections and records under Assistants menu, matching Skills admin convention.

• 🖥️ CLI Coverage Enhancements. Comprehensive WP-CLI command coverage improvements across the plugin toolchain.

• 📚 Folder README Convention — Unix Theory P7. Folder READMEs added for every PHP-bearing subdirectory across includes/ and addons/pro/includes/. Agent context docs (CLAUDE.md, AGENTS.md) synced with recent features.

• 🔧 Build & CI. build-spa-addons GitHub Actions workflow added. Missing SPA addon ZIPs restored. All SPA bundles rebuilt.

Privacy & Terms Notice: This plugin connects to external AI services. Review each provider's policies:

• OpenAI: Terms | Privacy

• Google Gemini: Terms | Privacy

• Anthropic: Terms | Privacy

• Cloudflare: Terms | Privacy

• Hugging Face: Terms | Privacy

• Ollama: Self-hosted (no external data transmission)

• OpenRouter: Terms | Privacy

• DigitalOcean Serverless Inference: Terms | Privacy

• DeepSeek: Terms | Privacy

• NVIDIA NIM: Terms | Privacy

• LM Studio: Self-hosted (no external data transmission)

• Baseten: Terms | Privacy

See the complete External Services Reference for all 20 services.

The plugin works standalone with ~303 base tools and optionally extends through the Pro addon, which adds ~1,263 Pro tools for advanced integrations (WooCommerce, JetEngine, social media APIs, GitHub, Google services — including Google Calendar and the new Gmail/Drive read tools — Shopify, QuickBooks Desktop, Yahoo Fantasy Sports, ESPN Fantasy, ECA management, CRE Debt & Securitization, Cloudways server management, CRM lead/deal/customer lifecycle, support ticket management, multichannel inbound/outbound messaging, Composio Connect, vision analysis object counting) and exec-based tools (FFmpeg, WP-CLI, Python rembg, Jukebox), bringing the total to ~1,566 built-in tools (~303 base + ~1,263 Pro; live count via WP_MCP_AI_Tool_Registry::get_tools() is authoritative).

Note on Tool Count: Tools include base WordPress operations, content management, media generation, research capabilities, and optional third-party integrations. The base version (~303 tools) works standalone. The full version requires the Pro addon and provides ~1,566 total tools including specialized toolkits for e-commerce, social media, analytics, document generation, vehicle estimation, image validation, JetEngine MCP, A2A agent delegation, CRE Debt & Securitization, Cloudways infrastructure management, CRM lead/deal/customer lifecycle + support tickets + multichannel, MCP Apps, Composio Connect, vision analysis, and more. Live count via WP_MCP_AI_Tool_Registry::get_tools() is authoritative.

Addon Ecosystem: NV oOS ships a growing family of 27 installable addons: Pro (addons/pro/ — ~1,247 additional tools), Chat SPA (addons/chat-spa/ — React chat replacement), Docs Hub (addons/docs-hub/ — in-site documentation SPA), SaaS Controller + Cloud Worker (addons/saas-controller/ + addons/cloud-worker/ — NV oOS Cloud control plane), Cloudways Dashboard (addons/cloudways-dashboard/ — Cloudways server management), Toolkit Shell / Canvas / Canvas Toolkit / Document Editor / Media Studio (addons/toolkit-shell/ etc. — Toolkit SPA Blueprint Tier A–D), Media Worker (addons/media-worker/ — Docker-based Node.js media sidecar, v3.2.0: multi-tenant shared worker mode, per-site provider keys, worker routing with local fallbacks, native /api/crawl/* endpoints + Crawl4AI facade), Graphify (addons/graphify/ — knowledge graph), Comic Reader (addons/comic-reader/ — CBR/CBZ/CB7/CBT reader), Funiq Bridge (addons/funiq-bridge/ — Payload-to-WordPress bridge with React SPA), Fleet Operator (addons/fleet-operator/ — scoped op_ operator credentials for MCP/A2A supervisor agents like Hermes), LibreChat (addons/librechat/ — code interpreter, speech, web search reranker), Schedule Anything Platform + SPA (addons/schedule-anything-platform/ + addons/schedule-anything-spa/ — SaaS booking with Stripe), Tenant Router (addons/tenant-router/ — multi-tenant routing), Page Agent (addons/page-agent/ — AI-powered browser page control copilot), Checkout API (addons/checkout-api/ — vendor-side Stripe checkout/licensing service for premium addons), Algorave, Cornerstone3D, Crocoblock DS, Embedded, Fantasy Football. Separate standalone plugins: NVOOS Content Graph (plugins/nvoos-content-graph/ — visual knowledge graph), NVOOS Content Graph AI (plugins/nvoos-content-graph-ai/ — AI providers + chat + RAG), NVOOS Content Graph AI Platform (plugins/nvoos-content-graph-ai-platform/ — agents, A2A, blueprints, skills). See docs/developer/addons/toolkit-spa-blueprint.md for the blueprint all SPA addons follow.

🎯 Mission: Modernizing Small to Medium Business Websites

NV oOS is specifically designed to help small to medium-sized businesses fast-track their outdated, stale, or insecure company websites to modern technology standards—without the need to add yet another wrapper around API calls. Instead, we're trying to peel back decades of API wrappers with the help of AI, providing:

• Direct AI Integration - No middleware required. Connect directly to OpenAI, Gemini, Anthropic, Hugging Face, Cloudflare Worker AI, Ollama, LM Studio, OpenRouter, and DeepSeek without custom development

• Security-First Architecture - Built-in protection against nefarious usage with active monitoring and prevention systems

• Enterprise-Grade Features - Access to capabilities typically requiring expensive custom development

• Compliance & Audit Tools - Comprehensive logging, rate limiting, and usage tracking built-in

• Zero Technical Debt - Modern codebase following WordPress standards, ready for current technology stacks

🛡️ Active Security Monitoring

NV oOS actively prevents and monitors against nefarious behavior. The plugin includes:

• Nefarious Usage Monitor - Real-time detection of suspicious patterns and automatic emergency shutdown capabilities【F:includes/class-wp-mcp-ai-nefarious-usage-monitor.php†L1-L676】

• Root Security Key - Optional emergency authentication layer to prevent unauthorized reactivation after security incidents【F:docs/features/security/root-security-key.md†L1-L511】

• Granular Capability Controls - Every tool and API endpoint enforces WordPress capabilities to prevent unauthorized access

• Rate Limiting - Built-in protection against abuse with configurable limits per user, model, and time period

• Comprehensive Audit Logging - Track all API calls, tool executions, and security events for compliance and forensic analysis

• Input Sanitization & Output Escaping - All user input sanitized, all output escaped following WordPress security best practices

This is not a tool for circumventing security or promoting bad practices. Every feature is designed with security, transparency, and responsible AI usage as core principles. The plugin actively works to stop and prevent misuse before it happens.

Latest audit: See docs/operations/compliance/SECURITY_AUDIT_2026_04.md — the published summary of the April 2026 security & compliance code review (no Critical findings; 5 High items, 3 Fixed and 2 Partially Fixed). Full deliverables under docs/project/audits/2026-04/.

WordPress.org compliance hardening (May 9, 2026): docs/operations/compliance/WORDPRESS_ORG_COMPLIANCE_2026_05_09.md — findings B3, B8, B10, B13, and production vendor remap all resolved.

⚠️ Warranty & Safe Use

We make every effort to keep NV oOS safe and secure — but by design, it can be destructive and resource-intensive when not properly configured.

NV oOS grants AI assistants access to powerful WordPress operations. The same capability that automates real work can cause irreversible harm if misconfigured:

• Destructive tools — bulk content deletion, user management, file writes, mass email, WP-CLI, direct database operations

• API billing exposure — uncapped AI provider calls can exhaust quotas and trigger unexpected charges

• Server resource exhaustion — concurrent agentic loops and SSE streams can saturate CPU/memory on shared hosting

Before going live: test on staging, take verified backups, apply least-privilege tool permissions, enable rate limiting, and review the system prompt of every public-facing assistant.

📄 Full details: WARRANTY.md — security commitment, "AS IS" disclaimer, destructive-operations table, resource-consumption guide, and mitigation checklist aligned with OWASP, NIST SP 800-53, ISO/IEC 27001, and the WordPress Plugin Developer Handbook.

Patent Pending

NV oOS is the subject of a pending patent application for its novel System and Method for Dynamic AI Orchestration Layer with Real-Time Capability Gating and Resource Budgeting.

Application Number: 19/410,504

The patent covers NV oOS's innovative approach to implementing sophisticated AI orchestration in WordPress's request-based PHP architecture—a platform not designed for real-time streaming, asynchronous operations, or persistent state management. This technical achievement enables enterprise-grade AI capabilities on WordPress by recreating event-driven behavior within PHP's synchronous execution model.

Key Innovations Covered:

• Dynamic resource budget allocation during streaming operations

• Capability-based access control for AI tool execution

• Registry-state-based scheduling in stateless environments

• Metrics-driven budget adjustment for real-time optimization

• Persistent-behavior illusion in request-based architectures

The orchestration layer makes NV oOS unique in the WordPress ecosystem by solving fundamental architectural limitations that prevent traditional WordPress plugins from supporting advanced AI features. See the System Architecture section below for technical details on how these innovations work together.

🏗 System Architecture

NV oOS implements a comprehensive orchestration layer for managing AI operations during real-time streaming events. The system architecture comprises:

• 15 language-model providers — OpenAI, Gemini, Anthropic, DeepSeek, OpenRouter, Baseten, Kimi (Moonshot), Z.AI (GLM), DigitalOcean, NVIDIA NIM, Cloudflare Worker AI, Ollama, LM Studio, Hugging Face, Flowhub

• ~1,566 tool classes (~303 base + ~1,263 Pro; live count via WP_MCP_AI_Tool_Registry::get_tools() is authoritative) registered through a singleton Tool Registry

• 36 REST controllers (16 base + 20 pro) under the mcp-ai/v1 namespace

• 64 service classes powering orchestration, budgets, and workflows

• 5 authentication methods — WordPress nonce, assistant credentials, mesh keys, Auth0 JWT, guest tokens

• Toolkit MCP servers — per-toolkit JSON-RPC 2.0 servers exposed under /wp-json/mcp-ai-pro/v1/mcp/{slug}; discoverable at /.well-known/mcp

• 8 inline-async-tick consumers — cooperative tick-lock pattern eliminates WP-Cron startup latency for background jobs (transcript mining, async tool executor, SaaS Apply, Crawl4AI, Docs Hub rebuild, Graphify reindex, Harness eval, Gemini Veo polling)

• 7 LLM Harness layers (+ 1 Pro) — opt-in epistemic layers A–H activated per-assistant via the LLM Harness metabox

• Orchestration Phases 1–7 — HITL approval queue, prompt-injection detector, structured output, OTel exporter, DAG builder, durable runs, triggers/webhooks, sub-agents

📖 For a detailed explanation of how NV oOS extends standard SSE and MCP protocols with novel orchestration features, see ORCHESTRATION-LAYER-ARCHITECTURE.md

Core Orchestration Layer: Overcoming PHP's Limitations

Critical Context: Most real-time AI streaming systems are built with Node.js, Python FastAPI, or Go — platforms designed for asynchronous, event-driven operations. These platforms natively support:

• Long-lived connections and persistent state

• Non-blocking I/O and parallel execution

• Event loops and asynchronous callbacks

• WebSocket protocols and SSE streaming

NV oOS achieves the same capabilities in PHP/WordPress — an environment fundamentally not designed for these patterns — through a sophisticated orchestration layer that creates a "persistent-behavior illusion":

1. Real-Time Budget Enforcement - Monitors token/memory usage during streaming, prevents exhaustion through predictive allocation

2. Capability-Based Tool Gating - WordPress role-based access control for AI tool execution

3. Predictive Optimization - Analyzes usage patterns to prevent resource overruns before they occur

4. Distributed Orchestration - Multi-provider support with policy-aware routing

5. Auditability & Compliance - Complete governance layer with logging and rate limiting

6. Cron-Based Task Orchestration - Extends orchestration to async operations with budget inheritance

Multi-Agent Orchestration Enhancement (DeepSeek V4-Inspired)

Added: January 2026 (v1.1.0)

Building upon the core orchestration layer, NV oOS now includes a sophisticated multi-agent coordination framework inspired by DeepSeek V4's orchestration patterns:

Key Components:

• Agent Role System - Four specialized roles (Planner, Executor, Critic, Specialist) with role-specific capabilities

• Team Composition - Automated team assembly based on task requirements and profession expertise

• Coordinated Workflows - Multi-step workflows with agent delegation, result aggregation, and validation

• Team CPT Integration - Persistent team configurations with orchestration modes (single/sequential/parallel/swarm)

• Profession-Based Discovery - 296 professions auto-assigned agent roles via intelligent seeding across 17 knowledge bases

Example Multi-Agent Workflow:

// 1. Compose research team (planner + executors + critic)
$orchestrator = new WP_MCP_AI_Agent_Team_Orchestrator();
$team = $orchestrator->compose_team( array( 'task_type' => 'research' ) );

// 2. Execute coordinated workflow
// Planner decomposes task → Executors research subtasks → 
// Communication service aggregates → Critic validates quality
$result = $orchestrator->execute_team_workflow( $team, $task, $context );

Documentation:

• See Multi-Agent Orchestration for complete technical details

• See DEEPSEEK-V4-README.md for documentation suite overview

• See DEEPSEEK-V4-USAGE-GUIDE.md for practical examples

Why This Architecture Is Novel: Overcoming PHP's Limitations

• Event loops and background workers

PHP/WordPress, by contrast, is fundamentally request-based:

• Every HTTP request spawns a new process that dies after responding

• I/O operations block execution

• No persistent memory between requests

• No native event loop or async coordination

NV oOS solves this by implementing an orchestration layer that creates a "persistent-behavior illusion" — effectively recreating Node.js's event loop behavior within WordPress's synchronous, request-based architecture. This architectural compensation is the system's core technical innovation:

This makes NV oOS patent-worthy as a technical workaround — it achieves sophisticated AI orchestration in an environment specifically not designed for such patterns. See ORCHESTRATION-LAYER-ARCHITECTURE.md for the complete technical analysis.

Computer-Implemented Resource Management

The system operates as a computer-implemented method executing on a processor with memory, performing:

1. Dynamic Resource Budget Allocation: The orchestration layer dynamically allocates token and memory budgets to tool execution requests based on real-time system capacity and operation requirements. The WP_MCP_AI_Resource_Manager continuously monitors server resources (PHP memory limits, execution time constraints) and automatically adjusts operational parameters.

2. Capability-Based Access Control: Tool execution endpoints enforce granular capability-based access controls. Each tool in the registry declares required WordPress capabilities, and the REST API controller validates user permissions before allowing execution. This ensures secure, policy-driven access to all operations.

3. Registry-State-Based Scheduling: The WP_MCP_AI_Tool_Registry maintains tool availability state and schedules execution based on policy constraints. Tools are loaded conditionally based on dependency availability, and execution is scheduled according to assistant configuration and user permissions.

4. Metrics-Driven Budget Adjustment: The system continuously monitors execution metrics (memory usage, API response times, token consumption) and adjusts resource budgets in response to prevent resource exhaustion and reduce latency. The WP_MCP_AI_Token_Budget_Manager implements safety margins and dynamic chunking to prevent API limit overruns.

System Components

The system comprises a processor and memory storing instructions that:

• Monitor real-time resource availability through PHP runtime introspection

• Enforce capability checks at REST endpoint boundaries

• Schedule tool execution through a centralized registry

• Adjust token and memory budgets based on detected system metrics

• Maintain operation logs for audit and optimization

This architecture is embodied in non-transitory computer-readable media (PHP source files) that, when executed by a web server processor, cause the system to perform the complete resource management workflow. The implementation prioritizes stability, security, and efficient resource utilization across diverse hosting environments.

Symfony Process Integration (December 2025)

NV oOS Pro addon integrates the Symfony Process component for secure external command execution. This modern framework replaces direct exec() calls in 6 Pro tools and 2 supporting services, providing:

• Enhanced Security: Proper argument escaping and command validation

• Timeout Management: Configurable timeouts with graceful handling

• Better Error Handling: Comprehensive exception catching and WordPress-friendly error reporting

• Process Control: Real-time output streaming and cancellation support

Migrated Tools & Services:

• FFmpeg operations (video frame extraction, metadata reading)

• Python rembg (background removal)

• WP-CLI execution

• Meta AI Jukebox (music generation)

• Supporting services for video and audio processing

The Process Service (WP_MCP_AI_Process_Service) provides WordPress-friendly wrappers with WP_Error integration, making external process execution consistent with WordPress coding standards.【F:includes/services/class-wp-mcp-ai-process-service.php†L1-L220】【F:docs/history/2025/implementations/symfony-phases/SYMFONY_PHASE2B_PROCESS_INTEGRATION.md†L1-L100】

🆕 Latest Updates (v1.1.71 — September 2026)

September 5, 2026 — Rate-Limit Unlock, September Model Catalog, Content Graph 1.0.4, Wave E2 Queue Layer

• 🔓 Rate-Limit Unlock + Fixed-Window Timer — PR #6322. The general REST request limiter (check_rate_limit()) now uses fixed-window accounting ({count, first_seen} transient payload, TTL never extended past window end, legacy integer transients normalized on read) and returns the remaining time in retry_after. When the limit trips, the new wp_mcp_ai_rest_request_rate_limit_exceeded action makes WP_MCP_AI_Restriction_Registry flag a rate_limit restriction (scope rest, auto-release at window end) — the block now appears in the Command Center Restrictions tab and Token Manager with the existing Lift button, and lifting deletes the wp_mcp_ai_rate_limit_user_{id} transient for a fresh window. Guest (IP-keyed) blocks are not user-attached and expire on their own (docs/features/security/user-restrictions.md).

• 🧠 MemPalace Wing-Scope Fix + Post-K16 Singles — PR #6327. WP_MCP_AI_Tool_Wake_Up_Context::matches_wake_filters() enforces wing/room exclusions (the Graphify graph anchors only boost scores — they never excluded out-of-scope memories, leaking cross-wing memories into wing-scoped blocks). Test-only isolation from CI run 91942465749: transcript repository mock restored unconditionally, shared tool-registry instance restored untouched (tearDown wiped one-shot bootstrap tools like the Pro OKF tools), Site Health connectivity tests unset the other provider keys, transcript-mining logging resets the settings cache.

• 🛒 Checkout API Pipeline + Salt Fix — PR #6315. build/nvoos-checkout-api-v<version>.zip joins the addon build, addons/checkout-api/tests joins the main PHPUnit gate, and the build workflow gains a tests job. Production fix: the token/crypto classes derive wp_salt( 'auth' ) . wp_salt( 'secure_auth' ) instead of reading AUTH_KEY . SECURE_AUTH_KEY directly (latent fatal on salt-less installs).

• 🧭 September 2026 Model Catalog — PR #6328. 217 → 228 models (catalog version 2026.07.14 → 2026.09.05): new gpt-5.6 family, gpt-6-astra preview, gpt-image-2, claude-opus-5, claude-fable-5.1, claude-mythos-5, gemini-3.6/3.7/3.8-flash, gemini-3.5-flash-lite, gemini-3.1-flash-image-preview, deepseek-v4-flash-vision-exp, kimi-k3, kimi-k2.7-code; retired deepseek-chat/-reasoner/-coder (API retired 2026-07-24), gemini-3.1-flash (sunset 2026-09-01), imagen-4 (shutdown 2026-08-17) with migration-map successors; pricing drift fixes (deepseek-v4-flash input 10×, Claude Opus inputs $15 → $5); gemini-2.5-flash-image sunset 2026-10-02, claude-sonnet-4-6 sunset cleared; defaults gemini-3.6-flash / gpt-image-2 / kimi-k3. ~45 derived files; lib/core/, lib/wordpress-adapter/, and the Content Graph AI plugin keep their own mirrored tracks.

• 🌐 Content Graph v1.0.4 Visual Experience — PR #6318. Theme engine (one token registry, PHP/JS byte-identical, WCAG ≥3:1 palettes), appearance settings tab, 24 SVG type glyphs, interactive legend/minimap/zoom cluster/fullscreen/keyboard, GET /edges REST route with relationship color families, export at 1–3×, [nvoos_graph] + block parity attributes, three visual filters, and a checkout-fallback redirect to the vendor product page.

• 🌐 Content Graph AI Blocks + Settings Shell — PRs #6316, #6317. assistant-selector, tools-grid, knowledge-base, assistant-builder blocks (the Build Assistant Prompt tab embeds the tools grid + KB); SettingsValidator/AiSection/SettingsRegistry settings shell (validate-then-sanitize, real validations on provider-selection + chat-settings).

• 🌐 Platform Wave E2 Queue Layer — PRs #6319–#6321, #6325. AsyncJobQueue (new nvoos_content_graph_ai_platform/async_job_executors filter), QueueManager, JobQueueManager (atomic SELECT … FOR UPDATE SKIP LOCKED), DeadLetterQueue — byte-identical ports completing the platform queue pipeline; drive-by fix extracts AgentCapabilityBoundaryHooks to its own PSR-4 file (standalone fatal).

• 🔗 Connectors Links — PR #6314. Providers-tab notice + API-key credential-source hint now use admin_url( 'options-connectors.php' ) (the core Connectors screen) instead of the unregistered admin.php?page=connectors.

• 📚 New Coding-Time Skill — PRs #6323, #6324. .agents/skills/mcp-ai-wpoos-updates/ codifies Track A (docs & release catch-up) + Track B (model catalog & config updates); skill count 53 → 54; YAML frontmatter fix follow-up.

• 📦 Versioning — bumped to 1.1.71 across all version-bearing files. Pro addon: 1.1.71. Media Worker: v3.2.0 (unchanged). nvoos-content-graph: 1.0.4 (was 1.0.3). nvoos-content-graph-ai: 1.0.4 (unchanged). nvoos-content-graph-ai-platform: 2.0.0 (unchanged). Checkout API: v0.1.0 (unchanged). Docs Hub addon: 0.4.2 (unchanged). Tool count: ~303 base + ~1,263 Pro (~1,566 total; live count via WP_MCP_AI_Tool_Registry::get_tools() is authoritative). Provider count: 15. Addon count: 27. Bundled skills: 74 base + 41 Pro. Coding-time agent skills: 54. Stale 1.1.68 + 1.1.69 build ZIPs removed.

🆕 Latest Updates (v1.1.70 — September 2026)

September 5, 2026 — exec-Disabled Host Hardening + Fifth PHPUnit Repair Wave

• 🛡️ Shell-Call Hardening — PR #6295. Disabled exec/proc_open no longer fatals the Pro Packages / Document Generation / Image Production / Media pages or the OCR service: wp_mcp_ai_check_nodejs_available() / wp_mcp_ai_get_nodejs_version() (new, addons/pro/includes/npm-integration-filters.php) fall back to WP_MCP_AI_Process_Service; OCR routes through is_cli_tool_available() / run_cli_command() with guarded proc_* calls; max( 0, (int) $value ) replaces absint() for assistant_id, time limits, and passing scores across seven CPT/settings pages.

• 🐛 Wave-5 Fixes — PRs #6280, #6281, #6282, #6283, #6285, #6286, #6293. Transcript display.attachments preserved (sanitizer kept the {url, label, downloadName, meta} shape) + model-aware max_completion_tokens caps (#6280); Pro dashboard get_monitoring_event_stats() accepts both slug styles (#6281); validated web_search profession tags (#6282); preset fixes — phantom validate_contact_data step removed, hook moved to schedule_data.hook (#6283); mesh URLs validated pre-sanitization (#6285); guest bucket for opt-in memory capture (#6286); curriculum exporter skips empty structures (#6293).

• 🐛 Wave-5 Fixes — PRs #6296, #6298, #6300. WP_MCP_AI_JetEngine_CCT::is_storage_available() physical-table probe + deleter/writer/tool gating + conversation_import_* preset additions (#6296); complexity routing restored — enhanced client passes the explicit model only, falling back to the site default only when routing is disabled (#6298); JET_ENGINE_VERSION requirement on JetEngine gates, transactional-DDL-safe cct_table_exists(), JFB id + flat params, cached wp_mcp_ai_pro_get_tool_map() feeding the token-manager listing (#6300).

• 🐛 Wave-5 Fixes — PRs #6301, #6302, #6303, #6304, #6305. Auth0 audience validated structurally (identifiers, not endpoints — the Auth0 domain keeps the SSRF guard) + GitHub OAuth wp_mcp_ai_github_oauth_redirect_terminate seam (#6301); skill-pack slug normalization + wp_mcp_ai_skill_pack_installed for empty packs + markup REST 400 annotations + filterable wp_mcp_ai_legacy_sse_enabled (#6302); clamp sanitizers + wp_mcp_ai_plugins_integration_redirect_terminate seam + section-ID/tab contract updates (#6303); is_elementor_editor_init() + static::is_available() (#6304); acting-user read_post checks + positive-int user validation + Auth0 structural audience check (#6305).

• 🐛 Wave-5 Fixes — PRs #6306, #6308, #6311. timed-out timeout needle + provider-prefix-stripped model limits (nvidia/nemotron-… → nemotron-3) + Ollama native tool-call preservation + css_to_xpath() predicates + published_date + wp_mcp_ai_missing_image (#6306); agentic * + token_db_optimized allowlist events + started_at_ms trace durations (#6308); K16 — sanitize_key-stable webhook IDs with dotted event names, settings-repository blob fallback (request guard / security manager / destructive-ops gate now see dashboard-saved settings), acting-user chart unfiltered_html, social-publish post_date reset, PSO keyword inflections, site-builder heading-tag allowlist, schema-verifier required independence (#6311).

• 🧪 PHPUnit Repair Campaign, Wave 5 — PRs #6280–#6312 (~29 test-only). Clusters closed from two triage runs (docs/developer/testing-docs/CI-TRIAGE-91006542428.md, CI-TRIAGE-91771001271.md): chat transcripts, charts/dashboards, professions/teams, presets/remote connections, Cloudflare, mesh, memory, CPT management, PayHere, Mubert, OpenAI Images, composer drift ×2, Content Graph AI, pro schedule AJAX, model routing, Gemini, JetEngine/Graphify stub isolation (tests/helpers/jetengine-stubs.php), OAuth/credentials, REST + skill packs (K15 upload isolation in #6312), admin/settings, Elementor, permission gates, HTTP/provider clients, orchestration budget, logging/events, K16 misc singles. Cluster board compacted in the mcp-ai-wpoos-test-suite skill (37 patterns unchanged).

• 📦 Versioning — bumped to 1.1.70 across all version-bearing files. Pro addon: 1.1.70. Media Worker: v3.2.0 (unchanged). nvoos-content-graph: 1.0.3 (unchanged). nvoos-content-graph-ai: 1.0.4 (unchanged). nvoos-content-graph-ai-platform: 2.0.0 (unchanged). Docs Hub addon: 0.4.2 (unchanged). Tool count: ~303 base + ~1,263 Pro (~1,566 total; live count via WP_MCP_AI_Tool_Registry::get_tools() is authoritative). Provider count: 15. Addon count: 27. Bundled skills: 74 base + 41 Pro. Coding-time agent skills: 53.

🆕 Latest Updates (v1.1.69 — September 2026)

September 4, 2026 — Vision Analysis Toolkit, tagDiv Admin Compat, DeepSeek Schema Fix, ZipSlip Guard Revival

• 👁️ Vision Analysis Toolkit — PR #6267. New Pro toolkit in addons/pro/includes/tools/vision-analysis/: analyze_image_objects (detection/vlm/hybrid modes, annotate=true GD bounding-box copy as a media attachment) + WP_MCP_AI_HF_Vision_Inference_Service::count_objects(), WP_MCP_AI_Vision_Count_Normalizer, WP_MCP_AI_Vision_VLM_Client, WP_MCP_AI_Vision_Annotator, and the NV oOS → Vision Analysis settings page (off by default, SSRF-guarded). bin/run-tests-docker.sh now composer-installs dev dependencies for production-only clones.

• 🐛 tagDiv Admin Compat & Metabox Crashes — PRs #6266, #6278. SiteKit tools return string capability-flag arrays (the undefined CAPABILITY_CAN_USE_IF_ADMIN constant killed the tools metabox at sitekit_get_adsense on nugl.com); the sortable fallback prints its bundled copy whenever td_wp_admin is enqueued (print-queue detection) at admin_head + wp_print_footer_scripts priority 1; the media script chain is forced as direct head tags from core-registered handles; tools metabox CSS enqueued on admin_enqueue_scripts.

• 🔧 DeepSeek Empty-Schema 400 — PR #6272. Argument-less tools now emit properties: {} across 29 files (Gmail/Drive list tools, Cloudways ×4, DietPi ×7, wait_for_user, okf_list_bundles, WebChat status, graphify, embedded abilities, ProfessionStatsTool, content-graph); LegacyToolAdapter preserves object maps + upgrades empty arrays; scaffold template emits new stdClass().

• 🔒 ZipSlip Guard Revival — PR #6270. count( $zip ) replaces the PHP 8.x-nonexistent $zip->num_files in all five entry loops (OKF bundle manager, Comic/Media Consolidate pages, Skill Manager ×2) — archive import rejects path traversal again (probe confirmed ../evil.md previously imported).

• 🐛 Fixes — PRs #6265, #6271, #6274, #6277. Rate limiter accepts the dispatching request's HTTP method (internal dispatches classified by their real verb, GET/HEAD exempt) and the nefarious monitor uses its own namespaced counter so a shared counter can't halve the chat budget; wp_mcp_ai_before_chat_request subscribers tolerate the legacy 2-arg emitter shape. Assistant edit screen: collect_prebuilt_shortcut_tasks() catches \Throwable per tool and coerces non-array filter results. Model catalog: gpt-4o → 128k, gemini-2.0-flash back in the video-capable list, claude-sonnet-4-6/gpt-4o restored to active, two Qwen Hugging Face entries added. Settings save reads the cache-suspension state before suspending so it restores correctly — WP 7.1 inline-async tick locks no longer jam after a settings save.

• 🧪 PHPUnit Repair Campaign Continuation — PRs #6260, #6262–#6276. Fourth wave (~9 test-only PRs): #6260 count-tokens params, #6262 continuation seam, #6263 chat SSE handler, #6264 deterministic cache expiration, #6268 coverage manifest + slug integrity, #6269 incorrect-usage notices (9 suites), #6273 admin registration drift, #6275 subtab sanitizer routing, #6276 14 Veo suites. The mcp-ai-wpoos-test-suite skill grew to 37 patterns; the plugin skill documents the third rate-limit counter.

• 📦 Versioning — bumped to 1.1.69 across all version-bearing files. Pro addon: 1.1.69. Media Worker: v3.2.0 (unchanged). nvoos-content-graph: 1.0.3 (unchanged). nvoos-content-graph-ai: 1.0.4 (unchanged). nvoos-content-graph-ai-platform: 2.0.0 (unchanged). Docs Hub addon: 0.4.2 (unchanged). Tool count: ~303 base + ~1,263 Pro (~1,566 total; live count via WP_MCP_AI_Tool_Registry::get_tools() is authoritative). Provider count: 15. Addon count: 27. Bundled skills: 74 base + 41 Pro. Coding-time agent skills: 53.

🆕 Latest Updates (v1.1.68 — September 2026)

September 3, 2026 — Pro SPA v2 Shortcode, Hermes Fleet Extensions, Stale-Nonce Self-Heal, Docs Hub 0.4.2, Provider Defaults

• 🧩 Pro SPA v2 Shortcode & Embedded Mode — PR #6256. [nvoos_pro_spa] embeds the Pro SPA v2 chat surface on the front end (chat-first embedded mode: threads, drawers, tool shortcuts, OKF drawer; router-free, no admin routes), with per-instance data-config attributes and optional guest mode through the base guest-token machinery behind the "Allow Guest Access" setting. Proposal 033 + addons/pro/README.md.

• 🖥️ Hermes Dashboard Fleet Extensions — PR #6249. New extensions/ tree: nv-oos-fleet fleet monitoring + control plane (Python plugin API, sites.yaml inventory, dashboard bundle), backup-download, external-app-tab, mcp-tool-shortcuts WebUI extensions, install.sh, and backend smoke tests. Plans in docs/developer/integration/.

• 🔄 Stale REST Nonce Self-Heal — PR #6225. GET /mcp-ai/v1/session/nonce mints a fresh session-bound nonce from the request's own auth cookie (nonce/capability-free by design, no-cache/no-store); the chat client retries failed rest_cookie_invalid_nonce 403s instead of showing "Cookie check failed" on full-page-cached or long-lived SPA pages. docs/reference/api/rest-api.md updated.

• 📚 Docs Hub 0.4.2 — PRs #6246, #6253. Local-page .md links rewrite to #/slug hash routes (anchors preserved); TOC slugs match github-slugger exactly (0 mismatches across ~63,000 headings); "Accept fix" computes targets relative to the source directory (case-preserving); ../ links validated with root-containment checks; broken-link entries carry slug/source_type + skip reasons; sync/rebuild reports "Atomic swap failed" when the staging cache is unwritable; dead 404 route removed; the core emoji loader is disabled on docs-browser pages (its MutationObserver was crashing React's DOM). Addon bump 0.4.1 → 0.4.2.

• 🔌 Provider Enable Defaults — PR #6255. Fresh installs now disable OpenAI/Anthropic/Gemini by default; provider dropdowns list only enabled + credentialed providers (get_available_providers()); the onboarding wizard auto-enables the provider whose key is entered; plugin + test-suite skills updated in-commit.

• 🐛 Fixes — PRs #6226, #6232, #6233, #6239, #6242, #6243, #6244, #6250, #6258. Calendar granted-scopes URL corruption (sanitizer keeps space-delimited scope lists; legacy %20 grants normalized back to spaces); agent-identity resolver casts alias-table canonical IDs to int; token-budget catalog dedup (gpt-4.1, gemini-2.5-flash) + restored wp_mcp_ai_model_tpm_limit filter seam on the CCT → catalog fallback path; assistant untrash restores the pre-trash status via wp_untrash_post_status (published assistants stay published, CCT sync resumes); tool presets gain the missing OKF enrich/router, Google Calendar, Gmail/Drive, composio_manage_accounts, and git_inspect/git_change tools; wp_mcp_ai_seed_task_templates AJAX action registered; fast-uri override >=4.1.4 across root/Pro/SaaS npm trees; Tiptap pinned to 3.30.4 in canvas-toolkit + document-editor (mergeAttributes prototype pollution); jQuery UI sortable shim on post edit screens for third-party admin scripts.

• 🧪 PHPUnit Repair Campaign Continuation — PRs #6223–#6257, #6259. Third wave (~21 test-only PRs: #6224, #6227–#6231, #6234–#6241, #6245, #6247, #6248, #6251, #6252, #6254, #6257 + CI alpha-log refresh #6223) kept the suite aligned with current contracts; the production fixes it carried are listed above, plus the late-merged wp_mcp_ai_attachment_segment_provider filter for providers without a remote file API (#6259).

• 🧹 Housekeeping. Stale 1.1.67 build ZIPs removed from build/, build/optional-components/, build/toolkit-addons/ (27 files).

• 📦 Versioning — bumped to 1.1.68 across all version-bearing files. Pro addon: 1.1.68. Media Worker: v3.2.0 (unchanged). nvoos-content-graph: 1.0.3 (unchanged). nvoos-content-graph-ai: 1.0.4 (unchanged). nvoos-content-graph-ai-platform: 2.0.0 (unchanged). Docs Hub addon: 0.4.2 (bumped this window). Tool count: ~303 base + ~1,262 Pro (~1,565 total; live count via WP_MCP_AI_Tool_Registry::get_tools() is authoritative). Provider count: 15. Addon count: 27. Bundled skills: 74 base + 41 Pro. Coding-time agent skills: 53.

🆕 Latest Updates (v1.1.67 — September 2026)

September 1–2, 2026 — Content Graph Platform Extraction v2.0.0, Ecosystem Port Wave D + D-UI, Google Workspace Read Tools

• 🧩 Content Graph Platform Extraction — PR #6123. plugins/nvoos-content-graph-ai-platform/ ships standalone at v2.0.0: Waves A–C + Blueprints (namespace-bridged admin UI, skill/slash-command/agent bridges, platform dashboard + settings registry, harness ToolRouterHarness), a 74-skill bundled-skills pack (data/bundled-skills/), knowledge base (data/knowledge-base/), and .github/workflows/phpunit-platform.yml. Plan: docs/project/plans/content-graph-platform-extraction-plan.md; gaps: MIGRATION-GAPS.md.

• 🧱 Ecosystem Port Wave D + D-UI — PR #6142. The additive Base+Pro → Content Graph port lands the AI runtime in nvoos-content-graph-ai — chat runtime core (D1), providers beyond the 13 (D2), model management + analytics/token tracking (D3), security guards (D4), plus D-UI admin pages, chat blocks, Elementor widgets, guest tokens, agent memory, WP-CLI, chat compat route, MCP JSON-RPC controller — with .github/workflows/phpunit-ai.yml. Content Graph AI bumps 1.0.3 → 1.0.4. Tracker: docs/project/ecosystem-port-tracker.md.

• 📧 Google Workspace Gmail + Drive Read Tools — PRs #6151, #6152. Six new Pro tools (Gmail: get_gmail_message, get_gmail_thread, list_gmail_connections, modify_gmail_message; Drive: get_drive_file, list_drive_connections) with two new clients on the shared includes/google/ foundation; search_gmail/search_drive updated.

• 🧪 PHPUnit Repair Campaign Continuation — PRs #6114–#6208, #6209–#6222. ~100 cluster PRs (second wave) kept the suite green; production fixes carried: assistant-directory REST empty-title guard (#6141), null-safe tool-error reporting (#6186), LLM sanitization delegated to the validator (#6187), display-metadata persistence (#6172), crawler job-contract hardening + wp_mcp_ai_crawl4ai_auto_spawn_cron filter (#6124, #6125), create-post taxonomy guards (#6115), toolkit-registry live-singleton resolution (#6117), slash-command handler fixes (#6122, #6140), URL-encoded external-API queries (#6156), Veo 5-second floor (#6180), token-tier caching guards (#6178), credential-resolver cache invalidation (#6119), cache-helper option-cache eviction (#6134), settings dashboard/registry fixes (#6139, #6144), orchestration-dashboard menu (#6160), tool-multiplier visibility (#6189), content-format filter seams (#6203), CRM workflow-preset canonical schema (#6208), Pro Composer class-ambiguity rename (#6198), PM assistant metabox (#6174), profession seeder envelope (#6197), destructive-ops gate canonical settings read (#6151), WhatsApp webhook signature rejection without an app secret (#6192), Graphify/Content Graph HTTP 304 cached-body re-serve (#6161), Veo async-job completion ordering (#6211), orchestration-renderer token-budget display (#6213), Shopify Sync client-availability filter seam (#6215), slash-command workflow metrics guard (#6216) + parallel/conditional block rendering (#6218), Paper Store array-field query matching (#6222), systemic PHPUnit 11 / WP 7.1 env-compat fixes (#6196). Test-suite skill refreshed to 26 patterns (#6154).

• 🧹 Housekeeping. Stale 1.1.66 build ZIPs + .sha256 removed; re-created 1.1.65 ZIPs re-removed.

• 📦 Versioning — bumped to 1.1.67 across all version-bearing files. Pro addon: 1.1.67. Media Worker: v3.2.0 (unchanged). nvoos-content-graph: 1.0.3 (unchanged). nvoos-content-graph-ai: 1.0.4 (bumped this window). nvoos-content-graph-ai-platform: 2.0.0 (new standalone). Tool count: ~303 base + ~1,262 Pro (~1,565 total; live count via WP_MCP_AI_Tool_Registry::get_tools() is authoritative). Provider count: 15. Addon count: 27. Bundled skills: 74 base + 41 Pro. Coding-time agent skills: 53.

🆕 Latest Updates (v1.1.66 — August 2026)

August 28–31, 2026 — PHPUnit Suite Repair Campaign, Test-Suite Skill, Content Graph AI 1.0.3

• ✅ New Coding-Time Skill — PR #6108. .agents/skills/mcp-ai-wpoos-test-suite/SKILL.md distills the campaign into an operational repair playbook: Docker test commands (WP 6.9/7.1), CI log-zip triage, 16 recurring root-cause patterns (hook resets, singleton interference, WP_Error envelope drift, nonce/user binding, WP 6.9 queue memoization, anonymous-class visibility), the production-fix-vs-test-fix decision, and cluster-PR branch/commit/validation conventions. Plugin skill updated with current contracts; AGENTS.md + copilot instructions registered; TEST-SUITE-REMAINING-FIXES-PLAN.md added. Coding-time skills: 52 → 53.

• ✅ PHPUnit Suite Repair Campaign — PRs #6008, #6009, #6012–#6107, #6109. ~100 cluster PRs brought the suite green. Production fixes carried by the campaign: assistant-access caching (wp_mcp_ai_assistant_access_cache_enabled filter + WP_Error caching, #6008); attachment-segment validation errors return explicit 400s (#6009); token-tier endpoint + tier-change audit logging (#6018); REST permission-callback allowlist refresh (#6019); bearer-auth context sync (#6014, #6016); REST analytics endpoints (#6015); job-queue closure serialization in legacy option storage (#6020); TIER_META_KEY chart-data constant (#6021); custom-table query guards against missing schema — Graphify DB/memory bridge, job store, tenant DB (#6022); restored WP_MCP_AI_Job_Notifier::update_status(), dot-preserving job IDs, source scoping (#6036); job-notifier REST routes for dot-IDs + owner auth (#6037); Little's-law metrics (#6038); job status promoted to running on progress updates (#6039); web-search result building restored for Exa/Perplexity (#6042); profession CPT meta clamping + primary-role flow (#6043); federation mesh-sync init (#6044); output-buffer leaks across render/AJAX paths (#6045); memory-capture failure envelope restored (#6049) + WPCS violations (#6052); container settings repository service ID (#6050); ecommerce toolkit enablement helper (#6051); A2A webhook logging + delegate tool errors (#6055); Media Toolkit tools normalized to WP_Error (#6057); autonomous-session tool error logging (#6058); Cloudways analytics adapter auth check (#6062); auto-categorize router/client (#6064); Graphic Editor Plus image access on WP 6.9+ (#6069); workflow-editor AJAX WP_Error/hook drift (#6070); Remove Background path guard + source resolution (#6072); quiz + document-template admin pages (#6089, #6090); profession service + aggregate results (#6096); NPM integration notice determinism (#6097); profession team CPT sanitization (#6105); Graphify admin classes load defensively on plugins_loaded (#6106); Shopify Sync CCT manager validates variant IDs before the JetEngine availability check (#6109).

• ✅ Content Graph AI 1.0.3 — PR #6056. Tool permission-check fix + addon bump; ZIP v1.0.3 ships in build/.

• ✅ Checkout API Addon + Content Graph Paid Checkout — PR #6063. New addons/checkout-api/ v0.1.0 (vendor-side Stripe checkout service: /wp-json/nvoos-checkout/v1/ session/verify endpoints, custom-table license store, signed expiring HMAC-SHA256 download URLs capped at 10/link, signature-verified idempotent webhooks with refund/dispute revocation and server-side license issuance on payment_intent.succeeded, per-IP rate limiting, storefront admin; Stripe secrets encrypted at rest, PHP 8.1+, vendor-server only). The free nvoos-content-graph plugin gains the client half: Stripe Payment Element modal, verify-via-vendor-API, local license key, one-flow install/activate from the signed URL, admin-only POST /payments/session|verify endpoints, four nvoos_content_graph/payments/* filters, throttled session creation, and download_url() + Plugin_Upgrader filesystem checks — no Stripe keys ship in the plugin. Contract: plugins/nvoos-content-graph/docs/commerce-vendor-api.md.

• 🧹 Housekeeping. Stale 1.1.65 build ZIPs + .sha256 removed from build/ (root, optional-components/, toolkit-addons/), matching the 1.1.63/1.1.64 cleanup convention.

• 📦 Versioning — bumped to 1.1.66 across all version-bearing files. Pro addon: 1.1.66. Media Worker: v3.2.0 (unchanged). nvoos-content-graph: 1.0.3 (unchanged; paid-checkout client merged, version bump pending). nvoos-content-graph-ai: 1.0.3 (bumped this window). Checkout API addon: v0.1.0 (new). Tool count: ~303 base + ~1,256 Pro (~1,559 total; live count via WP_MCP_AI_Tool_Registry::get_tools() is authoritative). Provider count: 15. Addon count: 27. Bundled skills: 74 base + 41 Pro. Coding-time agent skills: 53.

🆕 Latest Updates (v1.1.65 — August 2026)

August 26–28, 2026 — Reasoning-Model Fixes, Full-Crawl4AI Proxy, Security-Posture Closure, Chat/REST Hardening

• ✅ OpenAI Reasoning-Model Parameter Rejection — PR #5985. OpenAiCompatibleClient strips max_tokens/temperature from o-series/gpt-5 payloads (applyModelConstraints()) and retries 400 rejections with corrected payloads (sendWithParameterCorrection()) on both sync and streaming paths. Content Graph AI: EmbeddingService resolves the embedding provider deliberately (OpenAI preferred when keyed — most chat providers lack /embeddings) and honors the configured embeddings_model; \Throwable guards return WP_Error instead of fataling; the chat API reports graph_context_mode and falls back to keyword search over node labels without an index; the reindex button is always restored on server errors.

• ✅ Media Worker Full-Crawl4AI Proxy (031 Phase 3) — PR #5983. Env-gated POST /api/crawl/full + GET /api/crawl/full/task/:id in src/routes/crawl.js: every target URL is SSRF-validated before the payload is proxied to CRAWL4AI_FULL_URL, token-gated like every /api route, with 503 service_not_configured / 502 upstream_unreachable / 400 envelopes. TEMP_ROOT becomes the allowlisted sandbox root under STRICT_PATHS=1 (028 Q5, default flip deferred to worker 4.0.0). Worker stays v3.2.0. Proposals 028/031 updated.

• ✅ Security-Posture Findings Closed — PR #5981 (issue #5972). F-AI-01: Algorave Tone.js raw eval requires a per-session confirmation + warning banner (capability-scoped via nvoosAlgoraveConfig; Strudel unaffected). F-CMP-04: TMA markdown source map removed. F-AUTHZ-01: webhook __return_true callbacks carry justification comments. SECURITY_POSTURE.md updated in-commit.

• ✅ Chat, REST & Transcript Hardening — PRs #5991, #5993, #5994. Chat route no longer declares a top-level attachments arg (embedded segments authoritative; legacy clients tolerated); role enum validation moved to the sanitize layer (custom-role filter works again); attachment segment prep errors propagate to the client; orphaned tool messages silently discarded before dispatch; sign-preserving pagination sanitizer + guest-token helper; guardrail pre-screening skips array content.

• ✅ Webhook, Shortcode & Pro-Tool Fixes — PR #5990. Google Chat verification_token shared-secret auth when OIDC is disabled (?token= / X-Google-Chat-Token); Slack conversion anchor placeholders + italic-before-bold ordering; paper-store import/export missing-collection guards; idempotent scheduled-result block registration.

• ✅ Slash Commands & Blocks — PRs #6000, #5997. Toolkit manager re-resolves the handler at registration (init re-fire sync); CSV lists accept "1,2" and "1, 2"; assistant-builder/Pro toolkit blocks skip already-registered names (WP 7.1 notices).

• ✅ AJAX, TPM & Admin Settings — PRs #6004, #6005, #6006. Workflow-execution wp_send_json_* moved out of the try block; model-manager overwrite accepts 1/true/yes/on; research_model provider-error ordering; TPM falls back to the bundled model catalog (longest-prefix) when the CCT has no entry; dangling legacy high-token/security-monitor sections removed; WP 7.0 bridge checks WP_Connector_Registry::is_registered() first.

• ✅ Graphify & Remote-Connection Guidance — PRs #5986, #6003. Provider-prefixed *_key fields treated as sensitive; sync_remote_source returns canonical WP_Error; remote_wp_connection errors direct callers to list_connections first.

• ✅ Content Graph wp.org Refresh — PRs #5980, #5988, #5982, #5992. Icons (v5), main screenshot, generated page preview; Contributors field fixed; Content Graph AI chat tester fixed (settings store, graph adapter, SSE, core bridge).

• ✅ Test-Suite Cluster Fixes — PRs #5995–#5999, #6001, #6002. SSE stream-contract alignment, WP 7.1 icon-init replay neutralisation, transcript CI follow-ups, Phase 4 slash-command suites, SSE tool-result extraction, transcript roundtrip skip without CCT. Docs-only: doc open items → GitHub issues (PR #5979), Ralph Wiggum proposal recorded as implemented (PR #5984).

• 📦 Versioning — bumped to 1.1.65 across all version-bearing files. Pro addon: 1.1.65. Media Worker: v3.2.0 (unchanged). nvoos-content-graph: 1.0.3 (unchanged). Tool count: ~303 base + ~1,256 Pro (~1,559 total; live count via WP_MCP_AI_Tool_Registry::get_tools() is authoritative). Provider count: 15. Addon count: 26. Bundled skills: 74 base + 41 Pro. Coding-time agent skills: 52.

🆕 Latest Updates (v1.1.64 — August 2026)

August 24–26, 2026 — Google Calendar Connection, Composio Account Health, Log Hygiene, Non-Loggable Fields

• ✅ Google Calendar Connection & Shared Google Services — PR #5959 (32 files, +9,373/-206 lines). New includes/google/ foundation (OAuth service, Calendar v3 client, scope registry, credential resolver, sync + push) replaces four drifted Google OAuth copies; google_calendar connection type on both connection surfaces; six new Pro google-workspace tools (list_google_calendars, list_google_calendar_events, update_google_calendar_event, delete_google_calendar_event, check_google_calendar_availability, quick_add_google_calendar_event) plus a reworked create_google_calendar_event and a real sync_google_calendar. Scope enforcement on every write; push channels + Action Scheduler sync. Reference: docs/developer/architecture/integrations/google-calendar-connection.md, docs/reference/google-calendar-api-v3.md.

• ✅ Composio Verified Account Health & Real Tool Discovery — PR #5936 (23 files, +5,602/-258 lines). WP_MCP_AI_Composio_Account_Health ledger + live probe (catalog-discovered zero-argument read-only tools, curated fast path, honest status_only degradation), new seventh tool composio_manage_accounts (validate/reconnect/delete/prune, in-place reconnect), in-band failure conversion, v3.1 envelope unwrapping + corrected query params, ID-swap diagnosis by prefix, Health column with Verify/Reconnect in Remote Sites. See docs/composio-connect.md.

• ✅ Composio Proxied Provider Failures & Stale Account IDs — PR #5953 (11 files, +516/-35 lines). Proxied provider refusals inside data are now authoritative failures (401/403 → auth-required + reconnect machinery), bounded execution-error path walk, anchored status scan (content false-positives structurally excluded), stale connected_account_id fast-fails.

• ✅ Composio Zero-Argument Payload & Probe Determinism — PR #5958 (6 files, +677/-28 lines). Empty arguments encode as {} (never []), validation detail flattened, bounded upstream body on WP_Error, cache-index invalidation for filtered listings, deterministic probe ranking + curated Google Calendar probe.

• ✅ Composio Auth Config, Listings & App Removal — PRs #5932–#5934 (26 files, +1,715/-80 lines). Link-endpoint auth config resolution (auth_config_id/user_id/callback_url), real 401/403 error surfaces, API-key trim/validate; v3.1 listing envelope unwrap + Connected Apps list on the edit form; identity-bound execution (from_connection()), nonce-gated Remove that revokes the upstream grant.

• ✅ Tool-Declared Non-Loggable Result Fields — PR #5961 (16 files, +796/-15 lines). WP_MCP_AI_Tool_Sensitive_Result_Interface with dot-notation/*/subtree field declarations; the logger masks before preview truncation on the success, argument, and error paths; legacy-wrapper forwarding; additive-only filter; 8 declarers.

• ✅ Log Context Bloat & Rolling-Buffer Compaction — PR #5952 (21 files, +2,599/-22 lines). Per-entry byte budget on the persistence path (fingerprinted assistant_config/system_prompt, truncation, 12 diagnostic keys preserved), Extended budget 16→8 KB, previous_results no longer embedded (O(N²) fix), Data Management Compact/Delete via wp_mcp_ai_maintain_log_buffers.

• ✅ Secret URL Query Params in Logged Strings — PR #5954 (3 files, +1,238 lines). 26 credential-bearing parameter names masked in the shared redactor (scheme/host/path/non-secret params preserved) — covers tool arguments, error messages, and error_log(); ships the Google Calendar implementation plan doc.

• ✅ MCP JSON-RPC Error Paths & Diagnostics Re-wiring — PR #5957 (10 files, +372/-134 lines). /mcp jsonrpc/method no longer required so -32700/-32600 envelopes are reachable; MCP Server Diagnostics page, assets, and AJAX handlers re-wired after the entry-file rename.

• ✅ Validated-Tool Validation Restored on Symfony 5.4 — PR #5960 (14 files, +143/-107 lines). The enableAttributeMapping guard never fired on ^5.4 so constraint loading was silently skipped; version-appropriate helper + WPCapability constructor fix.

• ✅ Pro SPA v2 Conversation & Assistant Sync — PR #5962 (5 files, +537/-48 lines). selectSession carries assistant_id, stored sessions hydrate on mount, assistant switches start fresh conversations.

• ✅ Vision Tools Timeout Parameter — PR #5964 (20 files, +446/-113 lines). 5–300s timeout argument + global request_timeout inheritance via WP_MCP_AI_Vision_Request_Timeout; OCR client download/health paths covered.

• ✅ WP_Error Envelope Drift & Tool Bugs — PR #5965 (18 files, +203/-112 lines). Canonical-envelope test conformance, plan_schedules_from_workflow regex fix, double-prepare() fix, tracked job_id returns, cron pruning, model type constraint.

• ✅ Test-Env Registration & Suite Follow-Ups — PRs #5931, #5935 (66 files, +1,416/-564 lines). AJAX suite class loading, dispatch hardening, WP All Import bootstrap guard, moved-file requires.

• ✅ Content Graph wp.org Assets — PR #5963 (7 files). Banner/icon/screenshot assets for the nvoos-content-graph listing.

• 📦 Versioning — bumped to 1.1.64 across all version-bearing files. Pro addon: 1.1.64. Media Worker: v3.2.0 (unchanged). nvoos-content-graph: 1.0.3 (unchanged). Tool count: ~303 base + ~1,256 Pro (~1,559 total; live count via WP_MCP_AI_Tool_Registry::get_tools() is authoritative). Provider count: 15. Addon count: 26. Bundled skills: 74 base + 41 Pro. Coding-time agent skills: 52.

🆕 Latest Updates (v1.1.63 — August 2026)

August 22–23, 2026 — Artifact Evolution Phases A–G, Pro Addons Page, Storage Worker Offload, Test-Suite Exit Traps

• ✅ Artifact Evolution Phases A–G — PRs #5923, #5925 (52 files, +11,666/-75 lines). The Continual Harness Evolver + Meta-Harness become a complete gated Darwinian self-improvement loop for skills, prompts, and roles (proposal 007). Phase A repairs the broken evolve_harness ↔ Evolver contract (analyze_failures(), component-scoped evolve() with enforced dry_run, constructor normalization) and adds opt-in evolved-prompt consumption (WP_MCP_AI_Evolved_Prompt_Resolver) + evolved-skill merge into the Skill Registry, Refiner output sanitization/PII scrubbing, and a $5/hr default budget. Phases B–G add artifact-scoped eval suites with failure-case replay and post-mutation verification, artifact populations with parent sampling, learning-log-aware mutators, a pre-commit admission gate (three critics — contamination control), holdout-gated deployment with shadow A/B and drift rollback, and a governor with human approval queue + lineage graphs + an artifact-governance metabox on the assistant screen. New Settings → Orchestration Layer switches bridge the evolution filters. Every layer is opt-in and defaults off. 16 new/expanded test suites. Reference: docs/project/proposals/007-artifact-evolution.md + implementation plan; self-evolution governance section in docs/operations/compliance/EU_AI_ACT_2026.md.

• ✅ Pro Addons Admin Page — PR #5924 (7 files, +1,086/-1 lines). New registry-driven NV oOS Pro Dashboard → Addons page (/wp-admin/admin.php?page=wp-mcp-ai-addons, WP_MCP_AI_Addons_Page, 813 lines) installs and activates standalone addons whose ZIPs ship in build/ — nonce + install_plugins + allowlist gated; non-WordPress components listed read-only. AJAX test matrix included.

• ✅ Chat Storage Worker Offload (proposal 032 v2) — PR #5928 (30 files, +1,198/-210 lines). Conversation saves ≥ wp_mcp_ai_storage_worker_threshold (default 10,000 chars) offload JSON.stringify to the browser storage-worker.js via wpMcpAiStorageUtil; small saves, unload flushes, and worker failures fall back to the synchronous main-thread write with quota-retry. Per-call threshold overrides, storageWorkerUrl/storageWorkerThreshold localized on every chat surface, and a filter kill switch (0 = off). LLM worker manager URL resolution fixed (localized workerUrl → pluginUrl fallback → explicit error). Reference: docs/project/proposals/032-chat-web-workers-wiring-implementation-plan.md.

• ✅ DeepSeek Empty-Schema Fix — PR #5926 (6 files, +253/-17 lines). Schema normalization produced "properties": [] which DeepSeek rejects ("[] is not of type 'object'"); empty maps now encode as {} at all three payload boundaries (REST, Tool Service, ChatOrchestrator), and legacy tools are wrapped before the first register attempt so the fail-loud registry log no longer fires for successfully wrapped classes.

• ✅ OKF Skill-Knowledge Conformance — PR #5919 (91 SKILL.md files + 4 reference files). type: Skill frontmatter added to every bundled SKILL.md lacking it; reference files restored for wp-plugin-dto, wp-plugin-presenter, wp-action-scheduler, and mcp-builder; .agents/skills sync now recursive + covers all project-owned design skills. The OKF Validate tab no longer reports missing concepts.

• ✅ Test-Suite Exit Traps — PR #5929 (40 files, +2,121/-472 lines). New bin/sweep-tests.php (parallel 6-worker sweep, 180s/file cap, report + custom report path), AJAX test contracts in tests/bootstrap.php, and 24+ test files aligned. Production fixes kept minimal: bare-exit guards in four admin handlers (WP_MCP_AI_TESTS_RUNNING seams), duplicate CRM workflow-rule class renamed, toolkit MCP scope check now passes null (non-OAuth) scope and resource URIs skip esc_url_raw() (nvoos:// scheme preserved), SSE/Veo polling-limit filters added, export-provider log_action() public. Reference: docs/developer/testing-docs/TEST-SUITE-REMAINING-FIXES-PLAN.md.

• ✅ Pro SPA v2 Slash-Command Composer — PR #5920 (8 files, +520/-109 lines). /chat, /new, /clear, /copy, /export insert reliably from the SlashCommandsDrawer (also openable from the AgentPanel button); base assets/js/slash-commands.js fixes. Fix history: docs/history/2026/fixes/pro-spa-v2-slash-command-composer-fix.md.

• 📦 Versioning — bumped to 1.1.63 across all version-bearing files. Pro addon: 1.1.63. Media Worker: v3.2.0 (unchanged). nvoos-content-graph: 1.0.3 (unchanged). Tool count: ~303 base + ~1,249 Pro (~1,552 total; live count via WP_MCP_AI_Tool_Registry::get_tools() is authoritative). Provider count: 15. Addon count: 26. Bundled skills: 74 base + 41 Pro. Coding-time agent skills: 52.

🆕 Latest Updates (v1.1.62 — August 2026)

August 21–22, 2026 — OKF Bundle Management, Pro Skills Drawer, Vector Store Responses API Migration

• ✅ OKF Bundle Management & Pro Knowledge Routing — PR #5914 (47 files, +8,453/-284 lines). Ships the full OKF enhancement roadmap (Phases A–H): WP_MCP_AI_OKF_Bundle_Manager bundle lifecycle (create, rename, archive, delete, ZipSlip-safe ZIP import/export, health stats) with skill-knowledge protected; three new base tools (okf_list_bundles, okf_validate_bundle, okf_import_bundle) plus the okf_write_concept provenance schema — OKF tool surface now 10 tools; a new Bundle Manager admin screen (Bundles/Browser/Editor/Import-Export/Validate, nonce + capability-gated AJAX); Pro OKF-to-Skill Bridge (load_skill resolves bundle:concept_id with per-assistant grants + trust gating), auto-enrichment agent (okf_enrich_site_content), and hybrid knowledge router (route_knowledge_query). 10 new OKF test suites (116 tests) + 3 manual smoke scripts. Reference: docs/features/okf-integration.md, docs/project/plans/OKF-BUNDLE-MANAGEMENT-IMPLEMENTATION-PLAN.md.

• ✅ OKF Skills Drawer in Pro SPA v2 — PR #5915 (15 files, +3,148/-65 lines). New read-only mcp-ai-pro/v1/okf REST surface (bundle list/stats, concept browse + search, assistant skill grants) wired through the Pro module registry + SPA loader; new OkfDrawer component (1,020 lines) lets users browse OKF bundles/concepts and reference them from chat. Also declares WP_MCP_AI::$admin_okf_bundle_manager (PHP 8.2 dynamic-property fix).

• ✅ Percent-encoded OKF Concept Route Fix — PR #5916 (2 files, +52/-3 lines). Concept IDs containing / arrive as %2F and 404'd; the route pattern now allows % and the handler rawurldecode()s before the reader lookup.

• ✅ Vector Store Tools → Responses API — PR #5917 (10 files, +1,222/-73 lines). OpenAI removes the Assistants API on 2026-08-26: every vector-store call site drops the OpenAI-Beta: assistants=v2 header, file ingestion moves to file_batches with bounded polling (wp_mcp_ai_vector_store_batch_poll_max_seconds) and a headerless single-file fallback, and the clients gain filter / ranking_options support. New migration test suite (12 tests).

• 📦 Versioning — bumped to 1.1.62 across all version-bearing files. Pro addon: 1.1.62. Media Worker: v3.2.0 (unchanged). Tool count: ~303 base + ~1,249 Pro (~1,552 total; live count via WP_MCP_AI_Tool_Registry::get_tools() is authoritative). Provider count: 15. Addon count: 26. Bundled skills: 74 base + 41 Pro. Coding-time agent skills: 52.

🆕 Latest Updates (v1.1.61 — August 2026)

August 21, 2026 — Agent Identity Bridging, OKF Bundle Generator, undici Pin

• ✅ Agent Identity Bridging in Memory Store & Recall (29 files, +1,416/-165 lines). New WP_MCP_AI_Agent_Identity_Resolver (includes/services/class-wp-mcp-ai-agent-identity-resolver.php) canonicalises virtual agent keys (SPA drawer aliases, virtual_planner_*) to the canonical assistant post ID; the alias map (wp_mcp_ai_agent_id_aliases) is bounded (200), never autoloaded, and sanitised. store_agent_context resolves on store and echoes original_agent_id / agent_id_resolved; REST chat-memory recall merges alias buckets (stored_under per record, merged_sources, default limit 25); wing/room-scoped recall keeps single-bucket semantics. Drawers (base, chat-spa, pro-spa) show wing/room/stored-under chips, an agent-ID diagnostic, a show-all-scopes toggle, and store-triggered refresh; wake_up_context graph failures fall back to the transient path and scoped wake-up/recall errors retry unscoped. New tests: tests/test-agent-identity-resolver.php, memory-drawer JS/SPA additions. Reference: docs/features/memory/chat-client-integration.md.

• ✅ OKF Skill-Knowledge Bundle Generator — PR #5911 (6 files, +817/-1 lines). WP_MCP_AI_OKF_Skill_Knowledge_Generator generates the skill-knowledge bundle from includes/bundled-skills/ on bootstrap (priority 32) when missing or when the plugin version changed, and regenerates after admin bundled-skill reinstall — every okf_* tool now works out of the box instead of failing with "OKF bundle not found: skill-knowledge". Reference: docs/features/okf-integration.md.

• ✅ undici ^7.29.0 jsdom Compatibility — PR #5910 (14 files). jsdom 29.1.1 deep-requires lib/handler/wrap-handler.js (removed in undici 8), breaking vitest in seven addons; the override is pinned to ^7.29.0 — newest 7.x, all CVE fixes retained.

• ✅ nvoos-content-graph wp.org Review Reply & Report — PR #5912 (5 files, +260/-3 lines). WPORG-REVIEW-REPLY.md + WPORG-REVIEW-DETAILED-REPORT.md; WPORG-REVIEW-*.md excluded from distribution ZIPs.

• ✅ Content Graph CI ZIP Checksum Sync — PR #5906 (2 files). Build jobs upload and commit the matching ${ARTIFACT}.zip.sha256 (previously ${ARTIFACT}.sha256), ending checksum drift for CI-rebuilt ZIPs.

• 📦 Versioning — bumped to 1.1.61 across all version-bearing files. Pro addon: 1.1.61. Media Worker: v3.2.0 (unchanged). Tool count: ~300 base + ~1,247 Pro (~1,547 total; live count via WP_MCP_AI_Tool_Registry::get_tools() is authoritative). Provider count: 15. Addon count: 26. Bundled skills: 74 base + 41 Pro. Coding-time agent skills: 52.

🆕 Latest Updates (v1.1.60 — August 2026)

August 20–21, 2026 — Restricted-User Flagging, Conversation Import, Tool Schema Normalization

• ✅ Restricted-User Flagging & Unblocking — PR #5901 (26 files, +3,568/-20 lines). New WP_MCP_AI_Restriction_Registry (includes/class-wp-mcp-ai-restriction-registry.php) converts ephemeral rate-limit and token-budget blocks into persistent, enumerable restriction records (user meta + wp_mcp_ai_active_restrictions index, daily cleanup cron, audit-logged). The OOS rate-limiter adapter fires wp_mcp_ai_rate_limit_exceeded; chat rate limits are filterable (wp_mcp_ai_chat_rate_limit / wp_mcp_ai_chat_rate_limit_window). Admin surfaces: Token Manager "Restricted Users" panel (Base) + Pro Command Center Restrictions tab; REST GET /mcp-ai/v1/restrictions, GET|POST /mcp-ai/v1/users/{id}/restrictions, DELETE /mcp-ai/v1/users/{id}/restrictions/{type}; AJAX lift actions; wp mcp-ai restrictions list|lift|add; IETF rate-limit headers; new restriction_registry_on posture signal. Full reference: docs/features/security/user-restrictions.md.

• ✅ Conversation Import to Transcript CCT — PR #5898 (38 files, +8,791/-4 lines). New includes/conversation-import/ subsystem imports ChatGPT conversations.json (incl. ZIP), Google Takeout Gemini activity, Claude conversations.jsonl, ShareGPT, and OpenAI fine-tuning JSONL into the JetEngine ai_chat_transcripts CCT — one row per conversation — with format detection, tree/branch linearization, idempotent dedupe, dry-run, batching, and checkpoint resume. Four new tools (conversation_import_detect|run|status|delete), an admin upload/preview page with progress reporting, wp mcp-ai conversation-import detect|import|status|delete, GDPR exporter/eraser + retention coverage, media sideloading, and opt-in memory mining via mine_agent_memory. Five PHPUnit suites. Guide: docs/user-guides/conversation-import.md.

• ✅ Tool Schema Normalization — PR #5903 (5 files, +162/-14 lines). Tool argument schemas are normalized before being embedded in provider payloads — WP_MCP_AI_DeepSeek_Client, REST /tools schema output, WP_MCP_AI_Tool_Service, and the OOS ChatOrchestrator — preventing provider streaming failures from non-compliant schemas; the registry now logs tool registrations skipped for a missing tool contract.

• ✅ WP_Error Fatal Fixes — PR #5905 (8 files, +216/-50 lines). Memory tools (mine_agent_memory, retrieve_agent_memory, wake_up_context), the request guard, and the REST chat-memory controller now handle WP_Error returns before array/object access.

• ✅ nvoos-content-graph v1.0.3 wp.org Resubmission — PRs #5897, #5899, #5904. WPCS fixes for the renamed plugins, wp.org review fixes, readme/README polish, .wordpress-org screenshot assets, and the standalone build ZIP.

• 📦 Versioning — bumped to 1.1.60 across all version-bearing files. Pro addon: 1.1.60. Media Worker: v3.2.0 (unchanged). Tool count: ~300 base + ~1,247 Pro (~1,547 total; live count via WP_MCP_AI_Tool_Registry::get_tools() is authoritative). Provider count: 15. Addon count: 26. Bundled skills: 74 base + 41 Pro. Coding-time agent skills: 52.

🆕 Latest Updates (v1.1.59 — August 2026)

August 19, 2026 — Media Worker Crawling & Crawl4AI Facade, Research Tools Hardening, Docs Hub + Tool Registration Fixes

• ✅ Media Worker Crawling & Crawl4AI Facade — PR #5892 (25 files, +3,556/-16 lines). New /api/crawl/* endpoints in the worker — single-URL Markdown, batched crawling (sync or queued async), and link scans — using a two-tier extraction pipeline: static HTTP fetch → Readability → Turndown first (zero browser cost), with the hardened Chromium launcher as an automatic fallback for JS-heavy pages. Every URL passes the shared SSRF guard before fetch or navigation. A Crawl4AI-compatible facade (/api/crawl4ai/*) lets the plugin's run_crawl4ai_job remote mode target the worker as a drop-in Crawl4AI replacement, plus an LLM-based extraction utility (src/utils/llm-extract.js) for structured page data. Worker bumped to v3.2.0. Toolkit memory estimate now accounts for the worker sidecar (docs/features/TOOLKIT_MEMORY_TRACKING.md). Plan: 031-media-worker-crawl4ai-integration-plan.md.

• ✅ Research Tools Multi-Provider Hardening — PR #5893 (22 files, +2,728/-155 lines). semantic_content_search now resolves embeddings through the shared embedding-provider abstraction (OpenAI, Gemini, Ollama, DigitalOcean) independent of the assistant's chat provider, falls back to keyword search when unconfigured (fallback_mode: "keyword"), and skips stored vectors from mismatched models. New Gemini embedding provider (includes/services/embedding/class-wp-mcp-ai-embedding-provider-gemini.php). deep_research validates non-empty completions, falls back to reasoning_content for reasoning-only models, retries finish_reason: length with a doubled budget, walks the provider chain, and never caches empty reports. Two new read-only base tools: list_terms + list_taxonomies. New test suites: test-tool-deep-research.php, test-tool-semantic-content-search.php, test-tool-list-terms.php, test-tool-list-taxonomies.php, test-mempalace-phase3-embedding-providers.php.

• ✅ Docs Hub Rebuild & Broken-Link Suggestions — PR #5894 (9 files, +708/-54 lines). The copy-in-place updater bypasses Plugin_Upgrader, so upgrader_process_complete never fired and Docs Hub caches went stale after updates. The updater now fires wp_mcp_ai_plugin_updated after every successful in-place update; Docs Hub 0.4.1 reacts with a cache clear + async rebuild (cached remote content preserved) plus an admin_init version-mismatch guard for manual replacements/restores. Broken-link detection now resolves relative links against the slug map (fixing the "626 broken links" false positive on remote-only indexes); suggestions are case-insensitive with confidence clamped to [0.3, 1.0]. New tests: test-rebuild-job.php, test-indexer.php.

• ✅ Tool Registration & QA Infrastructure — PR #5895 (44 files, +1,151/-207 lines). Legacy-format tool classes (pre-interface) are transparently wrapped (WP_MCP_AI_Legacy_Tool_Wrapper + WP_MCP_AI_Tool_Legacy_Definition trait); ~32 previously-orphaned base tool files are now registered; a new wp_mcp_ai_tools_init action gives side-loaders (e.g. includes/orchestration-init.php) a late registration point; the registry tracks skipped tools in unavailable_tool_slugs. QA container memory limits raised (tests/qa/docker/docker-compose.qa.yml) and bin/install-test-plugins.sh hardened.

• 📦 Versioning — bumped to 1.1.59 across all version-bearing files. Pro addon: 1.1.59. Media Worker: v3.2.0. Tool count: ~300 base + ~1,243 Pro (~1,543 total; live count via WP_MCP_AI_Tool_Registry::get_tools() is authoritative). Provider count: 15. Addon count: 26. Bundled skills: 74 base + 41 Pro. Coding-time agent skills: 52.

🆕 Latest Updates (v1.1.58 — August 2026)

August 18, 2026 — Composio Connect, OOS Runtime Consolidation, Content Graph Rename

• ✅ Composio Connect Integration — PR #5889 (29 files, +5,318/-2 lines). New addons/pro/includes/composio/ subsystem bridges WordPress to the Composio MCP gateway: OAuth connection flow with state nonce (WP_MCP_AI_Composio_Auth_Handler), API client (WP_MCP_AI_Composio_Client), trigger bridge (WP_MCP_AI_Composio_Trigger_Bridge), and signed webhook controller (WP_MCP_AI_Composio_Webhook_Controller). Six new beta tools — composio_list_tools, composio_get_tool_schema, composio_list_connected_accounts, composio_create_connect_link, composio_execute_tool, composio_manage_triggers — plus remote-sites admin and metabox panels (connected-apps badge). 4 test classes (+887 lines). Full details: docs/composio-connect.md, docs/toolkits/composio-connect.md.

• ✅ OOS Runtime Consolidation Phases 0–5.8 — PR #5881 (8 commits). The OOS engine reaches parity with the legacy path: Phase 0–2 parity foundations (includes/bootstrap/oos-bridge.php, tool-surface + security-gate contracts), Phase 3 event-sourced session log (SessionLog/SessionEvent/SessionTelemetry + observer), Phase 4 shadow mode (includes/oos/ parallel parity runs serving the legacy result; wp mcp-ai oos parity CLI) + canary routing (per-assistant _wp_mcp_ai_engine opt-in), Phase 5 scoped tools (ToolScope) + compaction seam (CompactionProvider) + Pro composition & child binding (addons/pro/includes/composition/), and Phase 5.8 telemetry single-path. Full details: docs/project/proposals/029-oos-orchestration-runtime-consolidation-implementation-plan.md.

• ✅ Graphify Standalone Plugins → Content Graph — PR #5890. plugins/nvoos-graphify → nvoos-content-graph (v1.0.2), nvoos-graphify-ai → nvoos-content-graph-ai (v1.0.2), nvoos-graphify-ai-platform → nvoos-content-graph-ai-platform (v1.0.2), with matching build/sync workflow renames and bin/build-*-content-graph*.sh scripts. The addons/graphify/ knowledge-graph addon is unchanged.

• ✅ Security Center wp.apiRequest Fix — PR #5887 (1 file, +8 lines). The security tab's inline scripts (posture refresh, IP tests, snapshot restore, compliance export, self-test) call wp.apiRequest() but wp-api was never enqueued there. Now enqueued when ?tab=security is active. See docs/history/2026/fixes/security-center-wp-api-request-fix.md.

• ✅ deepmerge-ts CVE-2026-40345 — PR #5888 (4 files, +30/-6 lines). Override pinned in addons/media-worker/package.json and addons/pro/package.json.

• 📦 Versioning — bumped to 1.1.58 across all version-bearing files. Pro addon: 1.1.58. Media Worker: v3.0.0 (unchanged). Tool count: ~265 base + ~1,243 Pro (~1,508 total; live count via WP_MCP_AI_Tool_Registry::get_tools() is authoritative). Provider count: 15. Addon count: 26. Bundled skills: 74 base + 41 Pro. Coding-time agent skills: 51.

🆕 Latest Updates (v1.1.57 — August 2026)

August 15, 2026 — Updater Rework, Hermes Async Chat, Provider Detection Fix

• ✅ Plugin Updater: Base-Only Updates & In-Place Install — PR #5871 (5 files, +530/-266 lines). New base-only update flow for wp.org installs: check_for_base_update() plus the wp_mcp_ai_check_base_update / wp_mcp_ai_start_base_update AJAX handlers power a new "Base Version" panel in Settings → Advanced, resolving the GitHub base package (ASSET_BASE pattern, incl. legacy mcp-ai-wpoos-base-* naming) so base installs can update without upgrading to complete. Installation reworked from Plugin_Upgrader to copy-in-place (replace_plugin_from_zip()): the live plugin directory is never renamed, a backup snapshot is restored on failure, and base → complete upgrades deactivate a separately installed Pro addon to avoid double-loading. Pro version now read from the plugin header (get_pro_installed_version(), constant fallback), fixing drift in the imaging admin page and Pro CLI status. Full details: docs/features/plugin-updater.md, docs/history/2026/fixes/base-update-path-fix.md.

• ✅ Hermes WebUI Chat — Async Submit/Poll — PR #5872 (2 files, +15/-7 lines). hermes_chat submits via /api/chat/start and polls /api/chat/stream/status instead of a blocking request: runs keep executing server-side if the MCP client drops mid-run, HERMES_CHAT_TIMEOUT expiry returns status: "still_running" with the stream_id, and approval gates are answered with the configured HERMES_APPROVAL_MODE choice (ask leaves it pending). Answer extraction fixed for live WebUI payload shapes. bin/README.md updated.

• ✅ Fleet Operator Agent Context — PR #5873 (18 files, +574 lines). New addons/fleet-operator/.context/ tree: addon conventions, Hermes ops, MCP integration, design content, wp-plugin-dev rules, 6 operator roles (analyst/architect/developer/product-manager/qa/scrum-master), active/archive scratch areas, and task/memory templates.

• ✅ Service Status AI-Provider Detection — PR #5874 (2 files, +168/-20 lines). Provider detection now resolves credentials through WP_MCP_AI_Credential_Resolver (plugin settings, WP 7.0 Connectors, environment variables, PHP constants) with a merged-settings fallback for early bootstrap; Ollama remains keyless. New test suite tests/test-service-status-provider-detection.php. See docs/history/2026/fixes/service-status-provider-detection-fix.md.

• 📦 Versioning — bumped to 1.1.57 across all version-bearing files. Pro addon: 1.1.57. Media Worker: v3.0.0 (unchanged). Tool count: ~265 base + ~1,237 Pro (~1,502 total; live count via WP_MCP_AI_Tool_Registry::get_tools() is authoritative). Provider count: 15. Addon count: 26. Bundled skills: 74 base + 41 Pro. Coding-time agent skills: 51.

🆕 Latest Updates (v1.1.56 — August 2026)

August 14, 2026 — Media Worker Multi-Tenancy & Scale (v2.4.0 → v3.0.0), Worker Routing, Hermes Tooling

• ✅ Media Worker Multi-Tenant Shared Worker Mode — PR #5866 (21 files, +1,349/-304 lines). SITE_TOKENS JSON map switches the worker into per-site isolation: every /api/* request must carry a token belonging to a known site (fail-closed with AUTH_MODE=strict), and temp files, job queues, and rate limits are namespaced per site slug via the new src/utils/site-paths.js. Per-site rate-limit overrides (RATE_LIMIT_*_<SITE>), SITE_TOKENS_PREVIOUS rotation overlap, audit logs tagged site=<slug>. Specs: 026-media-worker-multi-tenancy-sidecar-proposal.md and 027-media-worker-multi-tenancy-phase2-spec.md. (PR #5866)

• ✅ Media Worker Phase 2 — Per-Site Provider Keys & Observability — PR #5868 (8 files, +366/-50 lines). SITE_PROVIDER_KEYS JSON map resolves per-tenant AI provider credentials before falling back to the shared pool (PROVIDER_KEYS_STRICT=1 disables the fallback); tenants.usage per-site provider usage counters; grouped temp TTLs (TEMP_TTL_UPLOAD/VIDEO/BROWSER/DOC) with storage stats; cluster-mode boot warnings and a k6 load-test kit under addons/media-worker/bin/load-test/. (PR #5868)

• ✅ Media Worker Phase 3 — Scale Without Breaking Changes (14 files, +602/-63 lines). Worker bumped to v3.0.0. W1: per-blog worker tokens on WordPress multisite (additive option chain in the sidecar trait). W2: new WP_MCP_AI_Media_Worker_Usage_Reporter (includes/class-wp-mcp-ai-media-worker-usage-reporter.php) — daily cron snapshots tenants.usage into an option and fires wp_mcp_ai_media_worker_usage_updated (opt-in, off by default). W3: SITE_TOKEN_<SLUG> / SITE_PROVIDER_KEYS_<SLUG> env-var merges over the JSON maps (env wins). W4: opt-in Redis rate-limit store (RATE_LIMIT_REDIS=1 + rate-limit-redis optional dep) for PM2 cluster mode. W5: PROVIDER_KEYS_FILE hot-reload with atomic replace and malformed-update rejection. W6 revised: permissive-mode boot notice only — the strict-path default flip is deferred (open Q5). W7: manual Velocity deploy workflow. 59 unit tests passing. Plan: 028-media-worker-phase3-proposal.md.

• ✅ Worker Routing Expansion (4 PRs, +2,370/-408 lines). Document generation (Word/Excel/PDF merge/watermark), OCR, video frame extraction, health charts, email (Nodemailer/MJML), QR/translate/PDF-extract, image vectorization, Prettier, and FFmpeg now route through the connected worker with local fallbacks unchanged; Pro settings page lists worker-routed packages. New addons/media-worker/bin/probe-wordpress.php probes the WordPress side (connectivity, tokens, package availability).

• ✅ Zero-Downtime Token Rotation & Canvas v3 — PRs #5864, #5865. WORKER_API_TOKEN_PREVIOUS accepted during the rotation overlap window; Canvas upgraded to v3 napi prebuilds for compiler-free installs on managed hosts. Cloudways readiness hardening (endpoint test script, queue/route resilience) — PR #5863. Live data/PDF route bug fixes — PR #5867.

• ✅ Hermes WebUI MCP Server, SSH Bridge & Skill Sync — PR #5862 + 3 follow-ups (8 files, +788/-13 lines). bin/hermes-mcp-server.js drives a Hermes agent from Zed as a stdio MCP server: serialized WebUI logins, session cookie + synchronous chat, tools hermes_list_sessions / hermes_chat / hermes_session_detail / hermes_sync_skills. bin/mcp-bridge-ssh.js + bin/utils/fake-ssh.js add the SSH bridge; bin/hermes-skill-sync.js / bin/sync-skills-to-hermes.js mirror the repo .agents/skills/ tree into the agent (idempotent upsert + prune); env-file parser hardened. Zed Hermes Console profile added to .zed/settings.json.

• 📦 Versioning — bumped to 1.1.56 across all version-bearing files. Pro addon: 1.1.56. Media Worker: v3.0.0. Tool count: ~265 base + ~1,237 Pro (~1,502 total; live count via WP_MCP_AI_Tool_Registry::get_tools() is authoritative). Provider count: 15. Addon count: 26. Bundled skills: 74 base + 41 Pro. Coding-time agent skills: 51.

🆕 Latest Updates (v1.1.55 — August 2026)

August 13, 2026 — MCP Agent Compatibility, Fleet Operator, Media Worker Hardening, Connection Pooling

• ✅ MCP Agent Compatibility & Tool Reliability — PR #5859 (8 files, +1,435/-32 lines). JSON-RPC errors now return HTTP 200 with the error envelope so client SDKs that drop non-2xx bodies relay tool errors instead of leaving agents hanging (auth/permission failures and pre-dispatch guards keep real HTTP statuses; filter wp_mcp_ai_mcp_error_http_status). Legacy MCP HTTP+SSE transport added for SSE-only clients (Accept: text/event-stream or ?stream=true) with a credential-bound session handshake from GET /mcp and responses delivered as event: message on the GET stream — backed by the new WP_MCP_AI_SSE_Session_Store (327 lines), gated by WP_MCP_AI_LEGACY_SSE_ENABLED. Tool rate limiter is now settings-driven (tool_rate_limit_max / tool_rate_limit_window / tool_rate_limit_exempt_tokens) with a security admin section; credential-token traffic exempt by default. GET/HEAD exempt from the request quota so discovery/SSE probe retry loops don't exhaust the hourly budget. Raw cred_* authorization headers (no Bearer prefix) accepted for agent configs that forward headers verbatim (filter wp_mcp_ai_accept_raw_credential_header). Async tool polling bounded at ~45s (15 polls × 3s) with inline kick of stuck jobs to stay under Cloudflare 524 timeouts. (PR #5859)

• ✅ Hermes Fleet Operator Addon — PR #5858 (18 files, +2,587/-8 lines). New addons/fleet-operator/ addon issuing scoped op_ operator credentials with audience binding, expiry, rate limits, and instant revocation so a supervisor agent (Hermes or any MCP/A2A host) can operate the site within an allowlist. Server-side MCP tools/list scoping plus tools/call enforcement and audit attribution via base-plugin hooks; /mcp method enum relaxed so ping and notifications from standard clients reach the JSON-RPC handler. Ships admin page, WP-CLI commands, Hermes config generator, 3-skill nvoos pack (nvoos-approvals, nvoos-operations, nvoos-site-context), runbook (docs/operations/fleet/hermes-operator-setup.md), and PHPUnit coverage (406 lines). (PR #5858)

• ✅ Media Worker v2.2.0 Security Hardening — PR #5857 (22 files, +1,247/-215 lines). Nine security controls: timing-safe X-Site-Token auth middleware, SSRF guard with OWASP controls (src/utils/safe-url.js, 248 lines), sandboxed Puppeteer launcher, express-rate-limit global + per-route-group limits, Helmet headers, split health endpoints with real capability detection, structured request logging (no secrets), graceful shutdown, and unit tests. Bumped to v2.2.0 with helmet ^8 and express-rate-limit ^7.5. Plugin sidecar: WP_MEDIA_WORKER_TOKEN constant now takes priority over the option; settings page shows constant mode, warns on cleartext URLs, and links the Velocity guide. New Cloudways Velocity setup guide (docs/operations/deployment/media-worker-velocity-setup.md), .env.example, and CI workflow. (PR #5857)

• ✅ Media Worker Subtree Sync — PR #5856 (4 files, +56/-2 lines). New sync-media-worker.yml GitHub Action mirrors addons/media-worker/ to the standalone nvdigitalsolutions/mcp-ai-wpoos-media-worker repo on pushes to main/alpha-working. Folder-level .gitignore keeps node_modules out of the standalone repo history; addon README doc links switched to absolute URLs that resolve after the split. (PR #5856)

• ✅ Database Connection Pooling Stance (Proposal 023) — PR #5855 (16 files, +2,572/-62 lines). Wave 1: Action Scheduler fallback enqueue gated when the RabbitMQ worker is active; concurrency guard slot tracking made atomic (Redis wp_cache_incr with InnoDB INSERT … ON DUPLICATE KEY UPDATE fallback); new mcp_ai_concurrency_slots table with daily cleanup cron. Wave 2: PDO persistent connection support in the Graphify remote SQL driver; DB polling cron gated behind the RabbitMQ transport preference; queue worker batch size configurable (--batch-size=N + filter); Site Health checks for MySQL pool, queue depth, and RabbitMQ health. RabbitMQ admin UI: section registered with settings registry, view added to Orchestration tab navigation, broken render() fixed. Full spec: docs/project/proposals/023-database-connection-pooling-stance.md. (PR #5855)

• ✅ PostCSS 8.5.26 — GHSA-6g55-p6wh-862q (1 file, +2/-2 lines). postcss minimum bumped to >=8.5.26 in addons/schedule-anything-spa/package.json. Version constraint only.

• 📦 Versioning — bumped to 1.1.55 across all version-bearing files. Pro addon: 1.1.55. Tool count: ~265 base + ~1,237 Pro (~1,502 total; live count via WP_MCP_AI_Tool_Registry::get_tools() is authoritative). Provider count: 15 first-class language-model providers. Addon count: 26. Bundled skills: 74 base + 41 Pro. Total coding-time agent skills: 51 (20 wp-* + 30 design-* + 1 mcp-ai-wpoos-plugin).

🆕 Latest Updates (v1.1.54 — August 2026)

August 12, 2026 — Security Fixes, Async MCP, Skill Enhancements, README Maintenance

• ✅ PostCSS CVE-2026-69153 — PR #5850 (2 files, +4/-4 lines). postcss minimum version bumped to 8.5.23 in package.json overrides across affected addons. The CVE (high severity) allowed crafted CSS to trigger unexpected behavior in the PostCSS plugin chain. No functional changes — version constraint only. (PR #5850)

• ✅ MCP Async Tool Response Handling — PR #5845 (1 file, +111 lines). Fixed tools/call endpoint dropping async tool responses. The handler was not waiting for deferred/promise-based tool execution results, causing silent failures for long-running tools (video generation, deep research, batch OCR). Now correctly awaits and serializes async results. (PR #5845)

• ✅ Plugin Updater Integrity Check v2 — PR #5846 (1 file, +16/-2 lines). Follow-up hardening to the v1.1.52 post-install integrity check: fixed a false-positive on the bridge/ directory path caused by stale PHP stat cache, added clearstatcache() after every file check to force fresh filesystem reads, and handled a main-file rename edge case where the old filename persisted in cache. (PR #5846)

• ✅ API Key Merged-Settings Fix for Research Tools — PR #5852 (20 files, +37/-33 lines). deep_research, web_search, semantic_content_search, semantic_context_search, retrieve_agent_memory, recall_memory, wake_up_context, and 13 other tools were reading API keys from get_option('wp_mcp_ai_api_key') directly instead of using the merged credentials system. This meant per-assistant and per-provider API key overrides were silently ignored for all research operations. All 20 tools now use WP_MCP_AI_Credentials_Manager::get_merged_credentials(), which respects the provider → assistant → global fallback chain. (PR #5852)

• ✅ Design Skills Audit & Enhancement — PR #5847 (61 files, +7,971/-57 lines). Comprehensive enhancement of all 22 existing design-* coding-time agent skills with detailed operational instructions, multi-step workflow guidance, cross-references to related skills, and specific MCP tool-calling patterns. 7 new pro-toolkit skills created:

  • design-ai-assistant-admin — manage AI assistants, model configs, provider setup, peer mesh networking

  • design-crm — complete CRM pipeline (leads, deals, companies, contacts, BANT/MEDDIC scoring)

  • design-project-management — projects, tasks, sprints, events, task plans, PM workflows

  • design-communications — multi-channel messaging (SMS, email, WhatsApp), contacts, delivery tracking

  • design-services — service components, dependency tracking, approval workflows

  • design-team-management — organizational teams, professions, role assignments

  • design-vault — encrypted vault for passwords, secure notes, payment cards, digital identities

  • design-security-ops — security operations, audit logging, posture management Total coding-time skills: 44 (22 wp-* + 22 design-*). (PR #5847)

• ✅ OKF YAML Frontmatter Compliance — PRs #5844, #5849 (65 files combined). Added missing type: Skill frontmatter field to all 22 design-* skills — required by the OKF v0.2 trust-signal specification for concept categorization (PR #5844, +68 lines in 45 files). Fixed YAML frontmatter spec compliance violations in 9 design skills where duplicate top-level keys and invalid field structures violated the OKF v0.2 schema (PR #5849, -100 lines in 20 files). All 44 skills now pass OKF v0.2 frontmatter validation.

• ✅ README TOC Anchor Fixes — PR #5853 (2 files). Fixed broken TOC anchor links caused by Unicode emoji rendering differences between GitHub's markdown parser and the anchor generator. VS16-based emojis (🪲🐛🛒📊) and U+26xx/U+27xx symbols (⚠️⚡✅) now generate correct, clickable anchor links in the Table of Contents. (PR #5853)

• ✅ Stale Build Artifacts Removal — PR #5851 (30 files). Removed stale v1.1.52 build ZIPs and compiled assets that were superseded by v1.1.53 builds. (PR #5851)

• 📦 Versioning — bumped to 1.1.54 across all version-bearing files. Pro addon: 1.1.54. Tool count: ~265 base + ~1,237 Pro (~1,502 total; live count via WP_MCP_AI_Tool_Registry::get_tools() is authoritative). Provider count: 15 first-class language-model providers. Addon count: 27. Bundled skills: 67 base. Total coding-time agent skills: 44 (22 wp-* + 22 design-*).

🆕 Latest Updates (v1.1.53 — August 2026)

August 12, 2026 — Shared Analytics Service, Circuit Breaker Hardening, Agent Skills Sync, SSE Fixes

• ✅ Shared Analytics Service — PR #5836 (18 files, +2,489 lines). New unified analytics subsystem (addons/pro/includes/analytics/) consumed by all NV oOS Pro toolkits. 7 platform adapters: Meta (Facebook + Instagram), Twitter/X, LinkedIn, TikTok, WooCommerce, Google Analytics 4, and Cloudways monitoring. 5 immutable DTOs (Account, Post, Metric, TimeSeries, Report) with strict coercion and WP_Error validation. Cross-platform normalization maps platform-specific metric names to a unified schema via WP_MCP_AI_Metric_Normalizer. Smart caching with transient-based per-data-type TTLs and cache stampede prevention via WP_MCP_AI_Analytics_Cache. Token-bucket rate limiter per platform via WP_MCP_AI_Analytics_Rate_Limiter prevents API exhaustion. Extensible adapter pattern — single WP_MCP_AI_Analytics_Adapter_Interface, register new platforms via $service->register_adapter(). MCP server registration in the Pro toolkit fleet. Site Health integration for analytics health checks. All 8 phases complete (DTOs → Adapters → Service → Tools → Migration → Admin UI → MCP Server → Site Health). Design Stack skill: design-analytics-reporting. (PR #5836)

• ✅ Circuit Breaker & Execution Pipeline Hardening — PR #5839 (12 files, +384/-42 lines). Circuit breaker protection added to all remaining AI provider clients that lacked it — now active on all 15 first-class providers with configurable failure thresholds and cooldown periods. Concurrency guard (WP_MCP_AI_Concurrency_Guard) wired into the execution pipeline to prevent overlapping destructive operations across the full tool execution lifecycle. Cost tracker (WP_MCP_AI_Cost_Tracker) per-operation cost estimation now enforced at the pipeline level, not just at the provider client boundary. Backpressure signals from concurrency guard and cost tracker now feed into the agentic loop's iteration budget, automatically throttling tool calls when system load is high. Plugin updater no longer shows base-only UI after upgrade to complete (base+Pro) package. Silent build asset failure now surfaces a clear error when the complete build asset is missing from a GitHub release. (PR #5839)

• ✅ Agent Skills Sync & Discovery — PR #5842 (1 file, +78/-24 lines). sync-agent-skills command now syncs project-owned design-* skills alongside the existing wp-* skills. The 22 new design-* coding-time agent skills (design-analytics-reporting through design-video-creation) are now auto-discovered by the Zed editor alongside the wp-* WordPress development skills. New mcp-ai-wpoos-plugin bundled skill covering complete NV oOS plugin operational guide (setup, assistant creation, MCP bridge tokens, API key bridging, Docker/WSL2 troubleshooting). Base bundled skills count increased from 45 to 67. (PR #5842)

• ✅ SSE Stream Backoff & Rate-Limit Fixes — PR #5841 (2 files, +18/-6 lines). SSE backoff counter now properly resets between successful stream reconnections, preventing unnecessary progressive delays. Rate-limit counter correctly cleaned up after stream termination, preventing false throttling on subsequent connections. (PR #5841)

• ✅ Load Guard Fatal Error Fix — PR #5840 (1 file, +3/-3 lines). Fixed WP_MCP_AI_Load_Guard calling private count_active_jobs() method, causing a fatal error on PHP 8.1+ when strict visibility enforcement is enabled. Now calls the public get_active_job_count() accessor instead. (PR #5840)

• 📚 Documentation Catch-Up (this release) — CHANGELOG v1.1.53 entry documenting Shared Analytics Service, circuit breaker hardening, skills sync, SSE/load guard fixes, and all changes from PRs #5836–#5842 that were missing from the v1.1.52 entry. README updated with v1.1.53 highlights and "What's New" section. CLAUDE.md updated: Paper Store section expanded with remote/REST/discovery details, bundled skills count corrected (45→67), circuit breaker/backpressure added to Security Infrastructure, Shared Analytics Service section added, context file references added for new docs. AGENTS.md updated: bundled skills count (45→67), coding-time skills count (22→44). Six .context/ files updated with August 2026 review dates and content: conventions.md (tool counts, analytics directory), tool-registry.md (list_mcp_tools, design-system preset), security-checklist.md (circuit breaker, post-install integrity check), rest-api.md (Paper Store REST endpoint), pro-vs-base.md, and .context/README.md.

• 📦 Versioning — bumped to 1.1.53 across all version-bearing files. Pro addon: 1.1.53. Tool count: ~265 base + ~1,237 Pro (~1,502 total; live count via WP_MCP_AI_Tool_Registry::get_tools() is authoritative). Provider count: 15 first-class language-model providers. Addon count: 27. Bundled skills: 67 base (up from 45).

🆕 Latest Updates (v1.1.52 — August 2026)

August 11, 2026 — Paper Store Remote, CPT Auto-Discovery, Design System Preset, Integrity Check

• ✅ Paper Store Remote Site Support — PR #5835 (13 files, +1,272/-73 lines). All 8 Paper Store tools (paper_store_list, paper_store_read, paper_store_search, paper_store_write, paper_store_update, paper_store_delete, paper_store_import, paper_store_export) now accept an optional connection_id parameter to proxy operations through Remote Site Manager. New REST API controller at mcp-ai/v1/paper-store (697 lines) with full CRUD + search + import/export endpoints for remote site access. New WP_MCP_AI_Paper_Store_Remote trait (96 lines) handles HTTP dispatch and response normalization. New list_mcp_tools discovery tool (234 lines) enables AI agent self-discovery — returns all available MCP tools with JSON Schema parameter definitions, filterable by toolkit namespace and search term. (PR #5835)

• ✅ Remote Connection CPT Auto-Discovery — PR #5834 (1 file, +16/-4 lines). Plugin-registered custom post types (including paper_store) now appear automatically in remote connection post type access controls via get_post_types(). Users no longer need to manually type each CPT slug, save, and reopen. resolve_post_type_access() mirrors the same discovery logic on form submission. Multi-Site Gateway plan (docs/project/plans/multi-site-gateway-plan.md, 1,429 lines) for hub-and-spoke federation targeting v1.5.0+. (PR #5834)

• ✅ Design System Tool Preset — PR #5837 (1 file, +108 lines). New design-system preset in WP_MCP_AI_Tool_Presets_Helper with 72 tools across 13 categories: WordPress content (9), image generation (4), image analysis (4), image processing (5), web search/research (5), charts (2), document generation (14), Paper Store (8), video generation (2), video processing (5), social media (3), AI/memory (6), and utility (4). (PR #5837)

• ✅ Post-Install Integrity Check — PR #5833 (1 file, +74 lines). WP_MCP_AI_Plugin_Updater now runs verify_installation_integrity() after every update, checking 15 critical file paths against the plugin directory before reporting success. Prevents silent update corruption on distributed filesystems (Cloudways) where unzip can drop files without errors. REST controller require_once calls now guarded with file_exists() checks. (PR #5833)

• ✅ Pro Update Visibility Fix — PR #5832 (1 file, +10 lines). Pro addon update section now visible for wp.org base + standalone Pro plugin installs.

• ✅ Docker Chmod Suppression — PR #5831 (3 files). Suppressed chmod() warnings during plugin install in Docker environments by checking filesystem constants.

• 📚 Documentation — 3 new feature docs: Paper Store, Remote Sites & Connections, MCP Protocol Version Negotiation. Updated: Tool Presets System (Design System preset), Plugin Updater (integrity check), DOCUMENTATION_INDEX, QUICK_REFERENCE, README, CHANGELOG.

• 📦 Versioning — bumped to 1.1.52 across all version-bearing files. Pro addon: 1.1.52. Tool count: ~265 base + ~1,237 Pro (~1,502 total; live count via WP_MCP_AI_Tool_Registry::get_tools() is authoritative). Provider count: 15 first-class language-model providers. Addon count: 27.

🆕 Latest Updates (v1.1.51 — August 2026)

August 11, 2026 — Orchestration Gap Remediation, MCP Version Negotiation, Media Worker Security Bumps

• ✅ Orchestration & Harness Layer Gap Remediation — PR #5830 (28 files, +3,295/-29 lines). Comprehensive audit of AI orchestration and LLM harnessing subsystem against August 2026 industry standards — OWASP LLM Top 10, EU AI Act, NIST AI RMF, MCP/A2A protocols, and OpenTelemetry GenAI conventions. OWASP LLM Top 10 coverage: 20% → 60%. EU AI Act compliance: 17% → 67%. Six new classes: WP_MCP_AI_Output_Guardrail (338 lines, OWASP LLM05 — validates LLM responses for PII, credentials, unsafe content before downstream consumption), WP_MCP_AI_Citation_Verifier (295 lines, OWASP LLM09 — cross-references claims against source documents to detect hallucinations), WP_MCP_AI_Model_Integrity_Verifier (274 lines, OWASP LLM03 — supply chain security with blocked model list and TLS verification), WP_MCP_AI_Semantic_Cache (289 lines, two-tier prompt caching: exact-match + embedding-based similarity), WP_MCP_AI_Canary_Deployment (404 lines, progressive model rollout with automatic rollback on threshold breach). Workflow Engine V2 graduated from feature-flag to GA release. Red Teaming scripts: bin/red-team.php (160 lines, 9 probe categories) and bin/red-team.sh (172 lines, 6 attack vectors). Harness init registers Output Guardrail + Citation Verifier by default. Three new docs: Orchestration & Harness Reference (202 lines covering all 10 layers A–J, protocols, security, observability, and 2026 standards alignment), EU AI Act Compliance Mapping (Articles 13–15, 50, 52), and OWASP LLM Top 10 coverage matrix. OWASP coverage: 20% → 60% | EU AI Act: 17% → 67% | Overall: ~65% → ~82%. All PHPCS checks pass with zero errors. (PR #5830)

• ✅ MCP Protocol Version Negotiation — PR #5829 (2 files, +123/-3 lines). Older MCP clients (Zed, Claude Desktop, Cursor) rejected the hardcoded 2026-07-28 protocol version. class-wp-mcp-ai-rest-mcp-methods.php (+77 lines) and class-wp-mcp-ai-toolkit-mcp-rest-controller.php (+49 lines) now negotiate the highest version both server and client support, falling back to 2024-11-05 when the client provides no version information. Maintains full backward compatibility. (PR #5829)

• ✅ Security Bumps — Media Worker — PR #5828 (2 files, +11,728/-3 lines). multer 1.4.5-lts.1 → 2.2.0 (deprecated 1.x line with multiple DoS vulns patched). nodemailer 8.0.5 → 9.0.5 (GHSA-p6gq-j5cr-w38f: raw header bypass fixed in 9.0.1). sharp 0.33.0 → 0.35.3 (libvips CVE-2026-33327 and related CVEs fixed in 0.35.0). No code changes required — all three packages maintain backward-compatible APIs across the version bumps. (PR #5828)

• 📦 Versioning — bumped to 1.1.51 across all version-bearing files. Pro addon: 1.1.51. Tool count: ~265 base + ~1,237 Pro (~1,502 total; live count via WP_MCP_AI_Tool_Registry::get_tools() is authoritative). Provider count: 15 first-class language-model providers. Addon count: 27.

🆕 Latest Updates (v1.1.50 — August 2026)

August 10, 2026 — Media Worker Sidecar, Site Health Fix, NPM Security, BMAD Agent Updates

• ✅ Media Worker Sidecar — PRs #5822, #5823 (132 files, +6,283/-2,779 lines). New addons/media-worker/ package providing a Docker-based Node.js sidecar that offloads heavy media processing from WordPress. Uses a service cascade pattern with Express routing across 11 handlers: browser (Puppeteer-based automation), code (sandboxed execution), data (transformation/ETL), document (generation/conversion), email (rendering/sending), image (manipulation/optimization), ocr (Tesseract-based), pdf (generation/manipulation), social (cross-platform posting), video (transcoding/extraction), and workflow (pipeline orchestration). Queue module (src/queue.js, 217 lines) with concurrent processing, retry logic, and Jest test suite. Pro integration: WP_MCP_AI_Media_Worker_Settings (256 lines) for admin configuration, WP_MCP_AI_Media_Worker_Client trait (219 lines) for tool-to-sidecar communication. Docker service name to private IP resolution in loopback detection (includes/class-wp-mcp-ai-http-helper.php). docker-compose.yml updated with media-worker service. Pro services updated (Fluent FFmpeg, Language Detection, MJML, Nodemailer, OCR, Prettier, Video Frame Extractor) and tools (export calendar ICS, analyze geospatial, generate health chart, render math equation). Documentation: deployment guide (280 lines) and proposal (237 lines).

• ✅ Site Health Redeclaration Fix — PR #5824 (1 file). Fixed PHP fatal error from function redeclaration in ensure_site_health_dependencies() inside includes/tools/class-wp-mcp-ai-tool-get-site-health.php. Now guarded with function_exists(). (PR #5824)

• ✅ NPM Security Fixes — PR #5821 (12 files). Resolved nanoid, js-yaml, and dompurify vulnerabilities across 11+ addon package-lock.json files (chat-spa, cloud-worker, comic-reader, docs-hub, document-editor, media-studio, pro/spa, pro/spa-v2, saas-controller, schedule-anything-spa, tenant-router, toolkit-shell). (PR #5821)

• ✅ BMAD Agent Editing Conventions — PR #5825 (2 files). Developer and QA agent YAMLs (.bmad/agents/nv-oos-developer.yaml, .bmad/agents/nv-oos-qa-engineer.yaml) updated with code editing conventions for consistent agent behavior during file modifications. (PR #5825)

• ✅ WPCS Formatting Cleanup — repository-wide. PHPCS fixes applied across all PHP files: tabs, docblock formatting, and code style corrections. No functional changes.

• 📦 Versioning — bumped to 1.1.50 across all version-bearing files. Pro addon: 1.1.50. Tool count: ~265 base + ~1,237 Pro (~1,502 total; live count via WP_MCP_AI_Tool_Registry::get_tools() is authoritative). Provider count: 15 first-class language-model providers. Addon count: 27.

🆕 Latest Updates (v1.1.49 — August 2026)

August 8, 2026 — Gemini Model Fix, Update Reactivation, Release ZIP Cleanup

• ✅ Gemini Model Resolution Fix — PR #5817 (1 file, 2 lines). includes/class-wp-mcp-ai-gemini-client.php — model resolution now uses the correct settings key instead of the deprecated fallback. The old gemini-pro model fallback has been removed; all Gemini model lookups now go through the canonical provider settings path. (PR #5817)

• ✅ Update Reactivation & Release ZIP Cleanup — PR #5816 (68 files, +164/-18 lines). includes/class-wp-mcp-ai-plugin-updater.php (+152/-18 lines) — update reactivation flow fixed; plugin now correctly handles post-update reactivation. Nonce-scoped update actions hardened. .distignore and .gitattributes updated to exclude development-only files (.agents/, .context/, .github/, IDE configs, test fixtures) from release ZIPs. bin/build-plugin-zip.sh updated with new exclusion rules. Minor fixes in OCR service, research report generator, research project tool, video analysis service, and 5 image-analysis tools. (PR #5816)

• 📦 Versioning — bumped to 1.1.49 across all version-bearing files. Pro addon: 1.1.28. Tool count: ~265 base + ~1,237 Pro (~1,502 total; live count via WP_MCP_AI_Tool_Registry::get_tools() is authoritative). Provider count: 15 first-class language-model providers. Addon count: 27.

August 8, 2026 — Shopify Sync Toolkit Fixes, PHPCS Security, Skill Catalogues (v1.1.48)

• ✅ Shopify Sync Toolkit — 7 Fixes (PR #5815, 6 files). Fatal error fix: graphql_query() → graphql() in Admin API connection test. Dead list_by_status code removed — replaced two broken per_page => 1 loops with a direct SELECT COUNT(*) WHERE status = %s. Always-zero analytics count now uses count($edges) from the actual API response. Duplicate get_option() call and redundant $sync_mode assignment removed. Infinite loop risk eliminated — paginated truncate() while-loop replaced with direct DELETE FROM {table} SQL. Double DB query per upsert halved — upsert() now accepts a by-reference &$operation output parameter. Missing orderby/order schema parameters added with sanitization passthrough.

• ✅ PHPCS CVE-2026-67434. squizlabs/php_codesniffer bumped 3.13.4 → 3.13.6 in all 3 composer.lock files. Resolves arbitrary code execution via crafted ruleset XML. Dev dependency only — no functional impact.

• ✅ Default Skill Catalogues. Brave Search Skills and all 22 WordPress Agent Skills added to default skill catalogue for new assistants out of the box.

• ✅ NVoOS Graphify Standalone Plugins. 3 new standalone plugin ZIPs in plugins/ directory: Graphify v1.0.1, Graphify AI Platform v1.0.0, and Graphify AI v1.0.0.

• 📦 Versioning — bumped to 1.1.48 across all version-bearing files. Pro addon: 1.1.28. Tool count: ~265 base + ~1,237 Pro (~1,502 total; live registry authoritative). Provider count: 15 first-class language-model providers. Addon count: 27.

August 7, 2026 — MySQL Connection Exhaustion Fix, Update Cache Bust, Mermaid Security (v1.1.47)

• ✅ MySQL Connection Exhaustion Fix (Cloudways) — PR #5809 (7 files, +438/-89 lines). Cron system overhaul in includes/bootstrap/cron.php — unified concurrency limits, per-process memory caps, and staggered offset scheduling prevent connection-pool saturation on Cloudways and similar hosts with restrictive MySQL connection limits. Each cron hook now respects a concurrency ceiling and memory budget before forking additional workers. Activation bootstrap (includes/bootstrap/activation.php) updated with resource-aware connection throttling — batched queries with explicit connection release during activation. Service status registry (includes/class-wp-mcp-ai-service-status-registry.php) hardened with transient caching to reduce database round-trips during status polling. Maintenance CPT streamlined. Pro status page: JS improvements (+99 lines), AJAX handler enhancements (+74 lines), dashboard fixes. (PR #5809)

• ✅ Update Checker Cache Bust — PR #5810 (1 file, +6 lines). ajax_check_update() and ajax_check_pro_update() in includes/class-wp-mcp-ai-plugin-updater.php now delete cached release data before fetching from the GitHub Releases API. Manual "Check for Updates" clicks always return live data instead of the stale 12-hour TTL cache.

• ✅ Mermaid Security Fixes (1 addon, 2 files). npm audit fix in addons/canvas-toolkit/: mermaid 11.15.0 → 11.16.1 resolves 5 CVEs (GHSA-3rrr-jr9j-h3q3 prototype pollution, GHSA-6x64-9x62-f2gx CSS injection, GHSA-rhh3-jpg6-66xh radar DoS, GHSA-2v8p-3f2j-5mp7 XY Chart DoS, GHSA-c4c3-pg64-4m4v config prototype pollution). Transitive bumps: dompurify 3.4.12 → 3.4.13, nanoid 3.3.16 → 3.3.18. Canvas toolkit dist bundle regenerated. Zero functional impact — mermaid mode is an unused stub.

• 📦 Versioning — bumped to 1.1.47 across all version-bearing files. Pro addon: 1.1.27. Tool count: ~265 base + ~1,237 Pro (~1,502 total; live count via WP_MCP_AI_Tool_Registry::get_tools() is authoritative). Provider count: 15 first-class language-model providers. Addon count: 27.

🆕 Latest Updates (v1.1.46 — August 2026)

August 5–6, 2026 — Backup & Restore, Plugin Updater, Abilities API, Status Page Fixes, PHPCS Cleanup

• ✅ Comprehensive Backup & Restore — Proposal 020 (14 files, +3,660 lines). New modular export system with export manager (512 lines) and 11 export providers (8 base + 3 Pro). Base providers: Core Settings (258 lines), Assistants (285 lines), CPTs (319 lines), Custom Tables (394 lines), Federation (253 lines), Addon Options (337 lines), Toolkit Options (282 lines). Pro providers: JetEngine CCTs (329 lines), License keys (190 lines), Remote Sites (324 lines). All implement WP_MCP_AI_Export_Provider_Interface with export() and import() methods. JSON-based export/import with chunked file I/O and progress reporting. Admin UI in Settings → Advanced expanded by 638 lines with provider checkboxes, export/import buttons, and progress feedback. Subsystem architecture documented in includes/admin/export/README.md (202 lines). Proposal 020 includes proposal (308 lines) and implementation plan (1,249 lines). (PRs #5805, #5807)

• ✅ GitHub-Based Plugin Updater — 5 PRs, +772 lines. New WP_MCP_AI_Plugin_Updater class fetches release metadata from the GitHub Releases API, compares versions against installed, downloads ZIP artifacts, and installs updates via WordPress Plugin_Upgrader. Base-to-complete upgrade path added to Settings → Advanced — one-click upgrade from base-only to complete package. Pro addon update support detects and updates the Pro addon alongside the base plugin when bundled; hides redundant Pro updater UI when already included. Core updates use Plugin_Upgrader; Pro updates use direct copy with rollback. Nonce-scoped update actions with proper capability checks. WPCS formatting and phpcs:ignore annotations for pre-existing nonce warnings in updater files. (PRs #5800–#5804)

• ✅ Abilities API Selective Adoption — Proposal 019 (11 files, +2,364 lines). New includes/abilities/ framework introducing machine-readable plugin operations with JSON Schema contracts for AI agent and MCP tool discovery. Registrar (188 lines) for ability registration/discovery/lifecycle. Bridge (235 lines) connecting abilities to the tool registry for MCP and AI agent discovery. Category Registrar (92 lines) for hierarchical ability grouping. Security Bridge (289 lines) providing capability-based access control per ability. New WP_MCP_AI_Tool_Ability_Interface (61 lines) — contract for tools exposing discoverable abilities. Bootstrap integration at abilities-init.php (49 lines). 5 PHPUnit test files: backward compatibility (139 lines), bridge (279 lines), registrar (190 lines), category registrar (96 lines), mock tool (158 lines). Reference documentation at docs/reference/abilities-registry.md (106 lines). Folder README at includes/abilities/README.md (33 lines). Proposal 019 includes proposal (220 lines), implementation plan (452 lines), and appendix (580 lines). OOS Bridge Wave 2 updated with new tool registry integrations. (PR #5799)

• ✅ Status Page Fixes — 2 files, +108 lines. Fatal error in Pro status REST endpoint resolved — missing methods in class-wp-mcp-ai-pro-status-ajax.php causing 500 errors on status API calls. JavaScript errors on the status dashboard page fixed with proper i18n text domain consistency. pro-status-page.js (32 lines changed) and pro-status-dashboard-page.php (12 lines changed). (PR #5806)

• ✅ Knowledge Base Auto-Build. CI workflow auto-builds knowledge-base.zip when playbook files change, ensuring the packaged knowledge base stays current with the profession registry. (PR #5797)

• ✅ PHPCS Cleanup & Canonical Envelope Compliance. Remaining parse errors from the canonical envelope conversion resolved across tool files. Text domain mismatches (mcp-ai-wpoos vs mcp-ai-wpoos-pro) corrected across Pro tool files. WPCS indentation, alignment, and inline comment fixes applied to 100+ files across base, Pro, and core tool directories. (PRs #5798, #5804)

• 📦 Versioning — bumped to 1.1.46 across all version-bearing files. Pro addon: 1.1.29. Tool count: ~265 base + ~1,237 Pro (~1,502 total; live count via WP_MCP_AI_Tool_Registry::get_tools() is authoritative). Provider count: 15 first-class language-model providers. Addon count: 27. Knowledge base: 311 professions (12 industry categories).

🆕 Previous Updates (v1.1.45 — August 2026)

August 5, 2026 — Self-Hosted OCR, Embedded v0.2.0, AI Transparency, Graphify, Build Automation

• ✅ Self-Hosted OCR (Unlimited-OCR + DeepSeek-OCR) — Proposal 018 (17 files, +4,087 lines). New unified vLLM client class supports both Baidu Unlimited-OCR (3B, MIT, 93.23% OmniDocBench) and DeepSeek-OCR (~3B, MIT) via self-hosted GPU infrastructure. Both models share the same integration pattern (vLLM + OpenAI-compatible REST API + NGramPerReqLogitsProcessor). New Pro tools: pro_unlimited_ocr (dedicated long-horizon OCR with structured output modes — text/structured/raw, table extraction, form field detection, Paper Store persistence) and pro_batch_ocr (Action Scheduler batch processing, sync up to 10 docs, async up to 100). Structured extraction service with marker parsing, table/form-field extraction. Existing extract_image_text (Base) and pro_document_ocr (Pro) tools now include unlimited_ocr and deepseek_ocr as provider enum values. Admin settings: two new endpoint URL fields with AJAX Test Connection buttons. OCR client: 640 lines.

• ✅ Embedded Addon v0.2.0. Backend registry expanded with voice tool calling, OpenMed healthcare tools, and new MCP — CCT Stability, Proposal 016+017 Hardening, API Key Fixes, Docs 🐛🛡️🔑📝

• ✅ CCT Stability Fixes (4 PRs). CCT duplicate registration race condition resolved with database-backed mutex lock (wp_mcp_ai_cct_registration_mutex option). FlowHub CCT duplicate registration corrected — second menu title entry removed, registration guard now checks for existing CCT before attempting re-registration. Base plugin fatal error when lib/core/ is absent (require_once of non-existent PlatformFlushInterface.php) — now guarded with file_exists() check, plugin continues gracefully without core features.

• ✅ API Key & Provider Fixes (2 PRs). Gemini API key resolution in video services (generate_video_via_veo, generate_video_via_imagen) restored after the v1.1.43 credential split moved keys to wp_mcp_ai_credentials option. Veo fallback chain now correctly forwards async context parameters (task ID, poll URL) between fallback attempts — fixes silent failures when primary model is unavailable. Omni API endpoint URL corrected. Audio completion hook registration fixed.

• ✅ Security & Architecture Hardening — Proposal 016 (all waves). 277 require_once calls for class-only files wrapped in spl_autoload_register conditional — reduces filesystem stat calls on every request. Autoload class-name heuristic corrected to strip wp_mcp_ai_ prefix from file names before resolution. Repository-wide phpcs auto-fix applied (trailing commas, blank lines, Yoda conditions, short array syntax). Docker QA container vendor mismatch fixed (runs composer install --no-dev inside container). All medium-severity findings addressed (error-log guarding, auto-index detection, debug output gate). 6 low-severity findings resolved (DNS info leak, master-key constant, version-drift alignment, shell-command denylist documentation, orphaned comment removal).

• ✅ Polling, Queue & Load Balancing Hardening — Proposal 017. Twelve structural weaknesses across three subsystems: polling timeout inconsistencies resolved with unified 30s cap; queue worker deduplication via wp_mcp_ai_queue_lock transient; load-balancer health-check interval standardized to 15s with jitter. Stale .php.bak backup file removed (was shadowing the real class file and causing ambiguous resolution).

• ✅ Deferred Security Items (#5755, items 5–7). Post meta values escaped via esc_html() before rendering in admin tables. Term description output hardened with wp_kses_post(). REST response field list filtered to remove internal-only debug keys (_debug_trace, _internal_state).

• ✅ Docs & Ecosystem. FOR_REVIEWERS.md updated for v1.1.43: total tool count ~1,500 (~263 base + ~1,232 Pro), 10 security classes, security posture status refreshed. 16 broken cross-reference links fixed across 5 docs files after the July docs reorganization. Graphify ecosystem audit completed: wp.org Plugin Check compliance verified, migration gap analysis documented, chat shortcode integration plan with NV oOS REST endpoints drafted.

• ✅ npm Security — Dependency Override Floors. undici >=8.10.0 (5 HIGH + 4 MODERATE CVEs) applied across 11 package.json files. fast-uri >=3.1.4 (CVE-2026-16221 backslash authority confusion) across 3 package.json files. ip-address >=10.4.0 (IPv4-mapped/NAT64 SSRF bypass + CIDR suppression) across 3 package.json files. All 11 package-lock.json files regenerated via npm install --package-lock-only.

• 📦 Versioning — bumped to 1.1.44 across all version-bearing files. Pro addon: 1.1.27. Tool count: ~263 base + ~1,232 Pro (~1,500 total; live count via WP_MCP_AI_Tool_Registry::get_tools() is authoritative). Provider count: 15 first-class language-model providers. Addon count: 27. Knowledge base: 311 professions (12 industry categories).

🆕 Previous Updates (v1.1.43 — August 2026)

July 30–August 1, 2026 — MCP Protocol Upgrade, Security Hardening, OKF v0.2, ICP System, Sync Fixes 🔒🆙📚🎯🔧

• ✅ Security Hardening v1.1.43 (16 files, 427 additions). SSRF protection across 7 provider connection handlers (blocks cloud metadata endpoints + private IPs). A2A agent URL SSRF guard with federation peer allowlist. SQL table-name regex validation for injection defense. A2A per-assistant agent cards now require authentication (was publicly exposed). Chat SPA /config and /manifest endpoints changed from __return_true to logged_in_permission. Missing args schemas on 7 REST endpoints (voice, security center, chat transcripts, A2A webhook). Guest rate-limiting changed from shared counter to IP-based hashed keys (DoS vector closed). 3 tool capability mismatches corrected (check_site_security, delete_cron_job, purge_cache). Centralized defense-in-depth capability gate in Tool Registry. PHP object injection prevention via safe_unserialize helper with allowed_classes => false. Post meta/term output escaping hardened.

• ✅ MCP Protocol Upgrade (2026-07-28). Largest MCP revision since launch — stateless core removes sessions and initialize/initialized handshake (SEP-2567, SEP-2575). New server/discover RPC for capability probing. _meta per-request for protocol version, client identity, capabilities. Mcp-Method/Mcp-Name headers on Streamable HTTP (SEP-2243). TTL/cache-scope on tools/list responses (SEP-2549). Legacy client shim preserves backward compatibility. 10 files changed. SSE transport preserved (deprecated per SEP-2596, 12-month off-ramp).

• ✅ OKF v0.2 Trust-Signal Support. Indentation-aware recursive descent parser replacing flat frontmatter. Inline YAML mappings, nested object lists, flow sequences. Reader: get_trust_tier() (unverified/machine-confirmed/human-reviewed), is_stale(), extended search() with trust filters. Writer: validate_bundle() checks v0.2 fields. New tool: okf_validate_attestation. All 6 existing OKF tools surface trust signals. 15 files, 27 standalone smoke tests. Full backward compatibility with v0.1.

• ✅ Pro Module Registry (Architecture). wp_mcp_ai_pro_init() decomposed from ~625-line monolithic init function into a PSR-4 module registry. Per-module loading with dependency ordering. Graceful degradation when modules or their dependencies are unavailable. Fixes broken require paths that caused class not found fatals under certain load orders.

• ✅ ICP System (Pro CRM Phase G). Data-driven Ideal Customer Profile system with 7-dimension scoring engine (Firmographic 25%, Technographic 20%, Intent 15%, Engagement 15%, Buying Triggers 10%, Economic Outcome 10%, Negative Signals 5%) — fit+intent separation, behavioral decay, negative disqualifier scoring. 2 new MCP tools: compute_icp_score (score Company/Lead CPTs or manual data) and manage_icp_profile (CRUD via AI). Admin UI with profile list table and 7-tab weighted editor.

• ✅ Profession & Playbook Sync Fixes (7 PRs). Playbook: bulk sync silent failures now report {synced, errors} to admin UI. Force-regenerated playbook files no longer incorrectly deleted. Undefined array key orphaned_attachments in playbook stats fixed. Test Model page assistant dropdown restored. Profession: duplicate slug detection in knowledge base JSON files (312→311 entries). Title fallback matching prevents Update from creating duplicate posts when slugs change between versions. Consumed-post collision fix ensures all unique KB entries create distinct posts (resolves 310-published vs 311-KB mismatch).

• ✅ Hexagonal Architecture Purity (PlatformFlushInterface). Last WordPress reference removed from nvoos/core library. New PlatformFlushInterface contract in lib/core/, WordPress adapter in includes/bridge/. SseHandler now receives flush interface via constructor injection — zero framework dependencies remain.

• ✅ Phase 3 Operational Security Hardening. Audit logger REST route hook corrected from init to rest_api_init. Security posture signal coverage hardened.

• ✅ Dependency Updates. wp-coding-standards/wpcs bumped to 3.4.1 for CVE-2026-45293. Dev tooling: replaced hardcoded Intelephense LSP with Phpactor in Zed settings.

• 📦 Versioning — bumped to 1.1.43 across all version-bearing files. Tool count: ~201 base + ~830+ Pro (~1,031+ total; live count via WP_MCP_AI_Tool_Registry::get_tools() is authoritative). Provider count: 15 first-class language-model providers. Addon count: 27. Knowledge base: 311 professions (12 industry categories).

🆕 Previous Updates (v1.1.42 — July 2026)

July 23–29, 2026 — Security Infrastructure, Framework-Agnostic Core, Status Page, Agent Skills, Algorave 🛡️🏗️📡🤖🎵

• ✅ Security Infrastructure Hardening (4 PRs). 7 new security infrastructure classes: Request Guard (SSE limits, JSON depth, body size, error verbosity, asset version stripping), Security Posture (21 signals, 0-100 weighted score, A-F grading), Destructive Ops Gate (confirmation gate), URL Guard, Concurrency Guard, Cost Tracker, API Key Store. Site Health checks for cron + security posture. 13 new security unit tests covering API key encryption, auth split-brain, break-glass, credentials expiry, destructive ops, rate limiting, SSE auth/CORS/rate limiting, SSRF protection, tool scope sanity, URL guard, and validated upload.

• ✅ Security Defaults & Headers Hardening. CORS configurable origin control (new cors_restricted posture signal, weight 5). Auth brute-force protection with rate limiting (new auth_brute_force_on signal, weight 4). Error verbosity three-tier filtering Safe/Moderate/Debug (new error_verbosity_safe signal, weight 4). Request body size enforcement (new body_size_limited signal, weight 3). OAuth token lifetime admin control, guest scope docs, SSE CORS filter, webhook secret indicator. DICOM PHI auto-redaction filter. Asset version stripping. Centralized exception guard.

• ✅ Security Dashboard Posture Signals. Dynamic signal count replaces hardcoded 17. Accurate audit-log empty-state text. CORS origin preview in header tool. Production hardening guide with WAF/OAuth/DICOM checklist.

• ✅ Framework-Agnostic Core (lib/core/). nvoos/core package with Hexagonal Architecture (Ports & Adapters), PHP 8.1+, separate composer.json. 32 domain contracts covering chat, caching, content, cost tracking, cron, email, events, files, HTTP, memory, models, OCR, providers, speech, streaming. 21 WordPress adapters. ChatOrchestrator with RateLimiter + SemanticCompressor. ProviderRouter (12-provider routing). Framework-agnostic ToolRegistry + SkillRegistry. 109 tools migrated to nvoos/core format. 5 chat parity gaps closed: finish_reason handling, prompt caching, input sanitization, vision image processing, transcript store contracts.

• ✅ Status Page & Incident Communication (Pro). Maintenance window CPT with REST CRUD, frontend banner + countdown timer, email/webhook/channel broadcast notifier. Incident CPT with phase state machine, phase-aware notification dispatcher, incident-to-lesson bridge. 4 new AI tools: get_service_status, create_incident, update_incident, resolve_incident.

• ✅ Agent Skills & BMAD Agents. 21 coding-time agent skills (.agents/skills/) for Zed editor covering WordPress plugin development patterns — abilities API, action scheduler, HTML API, i18n audit, plugin architecture, assets loading, bootstrap, cron, DTO, hooks, lifecycle, options storage, presenter, rewrite rules, query cache, REST API, security audit (with reference doc), security deep, security secrets, UTF-8 text. 6 BMAD workflow agent YAML definitions (.bmad/agents/) — analyst, architect, developer, product-manager, qa-engineer, scrum-master — with team composition config. Full .context/ subsystem context system with 8 topic files + 5 templates + active/archive directories.

• ✅ Algorave Addon. New standalone addon for live coding and algorithmic music generation. 9 tools: export-midi, generate-music-ai, generate-pattern, midi-output, modify-pattern, play-control, sample-manager, strudel-reference, visualizer. Pattern and session CPTs. REST API. Admin settings.

• ✅ Critical Bug Fixes. Request Guard wrap_dispatch parameter order mismatch with WP >= 6.5 rest_dispatch_request filter (was causing fatal error). Nonce authenticator now accepts _wpnonce query parameter in addition to X-WP-Nonce header (fixes 401 on legacy chat client streams).

• ✅ Dependency Security Updates. brace-expansion overrides across all addons. js-yaml and postcss security overrides. All npm lock files regenerated. undici pinned to 7.x for Node 20 compat in canvas-toolkit.

• 📦 Versioning — bumped to 1.1.42 across all version-bearing files. Tool count: ~201 base + ~830+ Pro (~1,031+ total; live count via WP_MCP_AI_Tool_Registry::get_tools() is authoritative). Provider count: 15 first-class language-model providers. Addon count: 27 (new: Algorave).

July 20–22, 2026 — OKF Integration, Security Compliance, Playbook Sync, Model Credentials, Dependency Bumps 📚🔒🔧🔑🛡️

• ✅ OKF Integration (Open Knowledge Format v0.1). Pure-PHP YAML frontmatter parser and bundle reader/writer with atomic concept CRUD and conformance validation. 6 new MCP tools: okf_read_concept, okf_browse, okf_traverse, okf_search, okf_write_concept, okf_delete_concept. All 41 bundled skills now OKF v0.1-conformant. Zero new Composer dependencies.

• ✅ Security Compliance (11 HIGH/P0 Fixes). HMAC-signed policy tokens for professional selector with 1h expiry. Health endpoint auth-gating (public response reduced to status only). ZIP path traversal prevention with realpath() containment. CSRF nonces on Shopify/EZuite/FlowHub sync endpoints. CDN SRI integrity hashes for all 6 libraries. Google Chat OIDC hardening with shared-secret token. Privacy imaging path traversal fix. Autoload fallback cleanup. Version alignment across all package.json files.

• ✅ Playbook Sync Fixes. Duplicate AJAX handler conflict resolved with nonce-aware cap bypass. CPT class loading guard in remove_duplicate_playbooks(). Silent sync failure detection — sync_all() now returns {synced, errors} surfaced to admin UI. Profession prompt regression fixed after HMAC refactor.

• ✅ Model Provider Credential Resolution. Model_Service now resolves API keys from all 4 sources — settings, credentials option, environment variables, and PHP constants — via Credential_Resolver.

• ✅ Dependency Security Bumps. adm-zip >=0.6.0 (ZIP CVE), axios >=1.18.0 (6 CVEs), brace-expansion 1.1.16/2.1.2/5.0.7 across 5 package.json files (CVE-2026-13149). npm audit: 0 vulnerabilities. composer audit: 0 advisories.

• 📦 Versioning — bumped to 1.1.41 across all version-bearing files. Tool count: ~201 base + ~830+ Pro (~1,031+ total; live count via WP_MCP_AI_Tool_Registry::get_tools() is authoritative). Provider count: 15 first-class language-model providers. Addon count: 26.

🆕 Previous Updates (v1.1.40 — July 2026)

July 12–15, 2026 — Content Format Awareness, Research Pipeline, Settings Credential Split, Model Catalog, Provider Parity, SSE Fixes

• ✅ Content Format Awareness. New WP_MCP_AI_Content_Format_Helper class (585 lines) detects and preserves Markdown, HTML, and plain text formats across post-modifying and analysis tools. Full 582-line test suite.

• ✅ Research → Paper Store → WordPress Draft Pipeline. All research tools support save_to_paper_store parameter. New create_post_from_research Pro tool. New action hooks. Backward-compatible.

• ✅ Settings Credential Split. Sensitive API keys moved to separate non-autoload wp_mcp_ai_credentials option with transparent merge via get_settings(). One-time migration. Defense-in-depth with wp_suspend_cache_addition. Settings import/export batch fixes (4 PRs).

• ✅ Demo Video Pipeline Complete (Phases 0–5). Scripted scene recording with AI voiceover. GitHub Actions CI workflow. 14 narration scripts.

• ✅ Kimi & DeepSeek Client Parity. Full streaming, tool use, and token tracking parity. DeepSeek and 9 missing providers added to all research tools.

• ✅ Model Catalog Update (July 2026). 24 files across base + pro. Default bumps: Gemini gemini-2.5-flash → gemini-3.5-flash, NVIDIA meta/llama-3.1-8b → nvidia/nemotron-3-nano-30b-a3b, Gemini Live gemini-2.5-flash-live → gemini-3.1-flash-live-preview.

• ✅ OOS Engine: SchemaStoreInterface + Tests. New domain contract, entities, tool, and WordPress adapter. 45 new unit and integration tests.

• ✅ SSE HTTP/2 Fixes. ob_clean() replaces ob_flush(). 524 timeout resolved in Pro SPA v2.

• ✅ Vector Store Sync No Polling. Status checked only on assistant change and page load.

• ✅ Validated Tool Slug Allowlist fix (PR #5680).

• 📦 Versioning — bumped to 1.1.40 across all version-bearing files. Tool count: ~195 base + ~830+ Pro (~1,025+ total; live count via WP_MCP_AI_Tool_Registry::get_tools() is authoritative). Provider count: 15 first-class language-model providers. Addon count: 26.

🆕 Previous Updates (v1.1.39 — July 2026)

July 10–13, 2026 — Meta-Harness Auto-Optimization, Agent Delegation Rework, Pro SPA v2 Polish, Tool Presets Refactor, CRM Enhancements 🧠🔄💬🎛️📊

• ✅ Meta-Harness Auto-Optimization System (All 7 Phases). New self-improving agent infrastructure: Trace Store + Trace Capture for execution telemetry (Phases 0-1); Harness Search Engine for telemetry querying (Phase 2); Pro Coding-Agent Proposer for automated improvement suggestions (Phase 3); Cues, Population, Auto-Deploy, and DSpark orchestration (Phases 4-6); comprehensive test coverage (Phase 7). Enables plugins to observe, analyze, and self-optimize AI agent execution.

• ✅ Agent Delegation — Major Rework. Delegation now runs inline instead of async for immediate results. REST-based dispatch via chat endpoint replaces no-op role executor. Cron resilience: delegation jobs no longer silently fail without retry. spawn_cron() added after wp_schedule_single_event for instant deferred job execution. Name-based agent resolution in delegate_to_agent tool. Wire-up and tasks drawer fixes in SPA v2.

• ✅ Pro SPA v2 — Polish & Fixes (20+ PRs). Vector store indicator and slash autocomplete positioning (PR #5666). Double path in vector store preload URL fixed (PR #5665). TDZ crash in CommandAutocomplete resolved (PR #5664). Inline command autocomplete and Zed-style refresh (PR #5663). Cost badges restored in response UI (PR #5661). allowSensitiveTools config propagated to chat-spa, Pro SPA v2, and delegation dispatch (PR #5658). Tool result rich rendering, sidebar auto-refresh, and media cache-busting (PR #5651). Auto-save transcripts on onFinish callback (PR #5650). Attachment support, save button, and storage display (PR #5646). Tasks drawer toolbar button with failedCount badge (PR #5642). Speech/audio tool response envelope fix (PR #5636). Capability flags rendering fix (PR #5635). Usage badges and image+text rendering fix (PR #5634). Sidebar media panel visibility and ID column (PR #5633). Media insert button + speech button reposition (PR #5632). System prompt leak, sidebar empty state, media-to-chat bridge (PR #5631). Tool display and conversation duplication fix (PR #5626). Composer mobile bottom padding (PR #5625). Media grid flexbox layout fix (PR #5624).

• ✅ Tool Presets Refactor. Essentials layers added, duplication stripped, auto-upgrade for validated variants (PR #5660). Validated tool auto-upgrade no longer causes duplicate names and not-allowed errors (PR #5662). Double tool execution fixed in SSE adapters with media refresh filter clear (PR #5652). tool_call_id fallback for DeepSeek streaming and SPA v2 tool results (PR #5638). Save button added, tool messages without tool_call_id stripped (PR #5648). Tool message block indentation auto-fixed (PR #5638).

• ✅ CRM Enhancements. CRM cache loop fixed (PR #5645). Upwork rate limiting (PR #5645). Freelance Platforms & External Sourcing configuration enhanced (PR #5639).

• ✅ Infrastructure. Veo 2.0 deprecated — Gemini Omni Flash replacement (PR #5667): migration path, deprecation detection preventing wasteful 404 fallback loops, user-facing error messages recommending gemini-omni-flash (10s duration, native audio, multi-turn editing), cost calculator updated, tool schema defaults updated, admin settings updated, tests updated. Workflow auth and protected method errors fixed (PR #5659). ZAP scan Docker network isolation resolved (PR #5637). npm packages rebuilt and built assets updated (PR #5653).

• ✅ Documentation. Comprehensive OpenMed integration plan v2 added (PR #5641).

• 📦 Versioning — bumped to 1.1.39 across mcp-ai-wpoos.php, WP_MCP_AI_VERSION constant (includes/bootstrap/constants.php), readme.txt, README.md, CHANGELOG.md, QUICK_REFERENCE.md, ROADMAP.md, and DOCUMENTATION_INDEX.md. Tool count: ~195 base + ~830+ Pro (~1,025+ total; live count via WP_MCP_AI_Tool_Registry::get_tools() is authoritative). Provider count: 15 first-class language-model providers. Addon count: 26.

🆕 Previous Updates (v1.1.38 — July 2026)

July 10, 2026 — Page Agent Addon, Pro SPA v2 Parity, User Memory Toggle, Tool Enhancements, Workflow Improvements 🤖💬👤📝🔄

• ✅ Page Agent Addon v0.1.0. New addon at addons/page-agent/ — AI-powered browser page control copilot powered by Alibaba Page Agent (MIT). Gives any WordPress page its own AI agent that can click, type, and navigate via natural language. Runs entirely client-side with no headless browser, Python, or Chrome extension required. Includes shortcode [mcp_ai_page_agent], Elementor widget, REST endpoints, MCP tool bridge, and admin settings page.

• ✅ Pro SPA v2 — Major Parity & Polish. Voice pipeline, tasks drawer, workflow tracker, and file attachment upload to WordPress Media Library. Tool Shortcuts and Slash Commands drawers for quick tool access. Mobile hamburger sidebar toggle. Speech/audio button fixes using correct ToolsClient REST endpoint. Conversation title improvements and turn count display fix. Button and token NaN fixes. Model sync and auth bypass fixes. Assistant preloading in runtime config to fix sidebar loading.

• ✅ Pro SPA v2 — UI Polish. Autoscroll fixes (scroll-to-bottom on submit, streaming start, user-at-bottom guard, restored via direct scrollTop instead of scrollIntoView). Viewport height fixes via CSS height chain instead of calc(100vh - Xpx). overflow:hidden on height-chain ancestors. filemtime-based cache-busting across all SPA addons. Lint errors resolved in both chat-SPA and pro-SPA. Deduplicated model selector to prevent React duplicate-key warning. REST route registration fixed and production assets rebuilt.

• ✅ Per-User Chat Memory Preferences. Users can now toggle chat memory on/off from their WordPress user profile. Individual control over AI memory retention without affecting site-wide defaults. Settings page integration.

• ✅ create_post / save_post Tool Enhancements. Markdown-to-HTML conversion via new WP_MCP_AI_Tool_Markdown_Converter trait — auto-converts Markdown content to HTML when creating or updating posts. Smart taxonomy suggestions auto-detect relevant categories and tags. Block content corruption fix for non-post post types.

• ✅ Workflow Blueprint & Schedule Improvements. Existing-content awareness in Content Publisher and Keyword Pipeline blueprints. Blog schedule presets now check for duplicate content before publishing. Readable response generation for workflow schedule result delivery.

• ✅ SPA Accessibility. Annotation pills made clickable with meaningful labels for screen readers.

• ✅ Security. OWASP ZAP DAST medium findings triaged as false positives.

• 📦 Versioning — bumped to 1.1.38 across mcp-ai-wpoos.php, WP_MCP_AI_VERSION constant (includes/bootstrap/constants.php), readme.txt, README.md, CHANGELOG.md, QUICK_REFERENCE.md, ROADMAP.md, and DOCUMENTATION_INDEX.md. Tool count: ~195 base + ~830+ Pro (~1,025+ total; live count via WP_MCP_AI_Tool_Registry::get_tools() is authoritative). Provider count: 15 first-class language-model providers. Addon count: 26.

🆕 Previous Updates (v1.1.37 — July 2026)

July 4–8, 2026 — JetEngine Meta Helper, Places Enrichment, RabbitMQ, Multi-Tenant DB, DSpark UI, Test Coverage, Docs Hub Broken Link Engine 🧩📍🐇🏢📊🧪🔗

• ✅ JetEngine Meta Helper (Universal). Unified REST exposure for all 25 Pro CPTs — show_in_rest and register_post_meta applied across the board. ECA _eca_* and _student_* fields registered in the JetEngine meta registry. MCP routing and error reporting bugs fixed. Load-order fix ensures the helper is available before dependent toolkit inits.

• ✅ Places Toolkit Enrichment. Two new AI tools — enrich_place_coordinates (batch geocoding) and enrich_place_details (Google Places API enrichment). Social media and booking fields added to the Places CCT schema. CLI bin scripts for batch enrichment operations. HTTrack import hardened: redirect stub handling, coordinate parsing, parent-child linking, and CLI parameter validation.

• ✅ RabbitMQ Client & Queue Infrastructure. RabbitMQ client wired with queue manager and async tool interception. Queue storage migrated from WordPress options to custom database tables for scalability. Health endpoint for queue monitoring. Dedicated queue worker for background processing.

• ✅ Multi-Tenant Database Isolation (Phase 0–4 Complete). Phase 0 laid the foundation with database isolation primitives and connection routing. Phases 1–4 delivered query scoping, schema isolation, cross-tenant safety guards, and admin controls — the full isolation stack is now operational.

• ✅ DSpark Admin UI & Speculative Orchestration. New admin page with settings, threshold configuration, efficiency dashboard, presets, and a hook system for extensibility. Speculative orchestration enhancements with configurable execution strategies.

• ✅ Crocoblock Design System Addon (All 5 Phases). New standalone addon at addons/crocoblock-ds/. Unified CSS custom properties, preset templates, and admin-controlled theming for JetEngine, JetSmartFilters, and JetFormBuilder. Registered in addon inventory and build pipeline.

• ✅ Test Coverage — 329 Tools Across 28 Toolkits. 301 previously untested tools now covered by batch tests. 6 previously untested pro toolkits with full coverage. 22 toolkits covered across 7 batch test files. HTTP testing infrastructure for remote tool tests. Docker plugin seed service for integration test coverage. CRM, PM, DietPi, and Cloudways test suites hardened with recursive file search, dependency resolution, and graceful skipping for abstract/missing classes.

• ✅ Docs Hub Broken Link Detection & Repair Engine. Automated broken link scanning with results table. One-click Accept fix buttons on the detail table. Suggestion engine regression fixed (was producing no suggestions).

• ✅ SPA Annotation Pills — Accessible. Clickable pills with meaningful labels for screen readers.

• ✅ CI/CD Hardening. OWASP ZAP DAST security scanning integrated into the CI pipeline. PHP bumped from 8.1 to 8.2 across all CI workflows. QA Docker setup fixed (unnecessary chmod on bind-mounted script).

• ✅ Webhook & Agentic-Loop Resilience. Proactive agentic-loop context compaction prevents context overflow during long-running agent sessions. Webhook Context Manager enhanced with industry best practices. Sliding window context loss fixed across all 11 webhook controllers. max_history_messages section default normalized to match base settings (8).

• ✅ Shopify Sync — Catalog API & Minimal Mode. Catalog API sync support for product caching. Minimal sync mode (title, SKU, stock levels only) for fast updates. Price field added to minimal mode payload.

• ✅ BME Chat History Context Strategy. RAG memory integration for chat history. Settings relocated to General → Behavior subtab on the new dashboard.

• ✅ Bug Fixes. Shopify Catalog API silent failures and result reporting. Recurring sync stuck at old interval with missing connection IDs. Sync log batching via deferred writes to avoid per-item DB timeouts. Webhook auto-reply broken by missing Webhook Context Manager require. Import-blueprints and remaining-pro-tools file discovery. Extended-cognition toolkit missing requires for trait and interface. Tool tests: static data providers, constructor mocks, slug overrides, WP_Error returns. js-yaml DoS vulnerability override. ESLint and WPCS compliance across test files and enrichment tools.

• 📦 Versioning — bumped to 1.1.37 across mcp-ai-wpoos.php, WP_MCP_AI_VERSION constant (includes/bootstrap/constants.php), readme.txt, README.md, CHANGELOG.md, QUICK_REFERENCE.md, ROADMAP.md, and DOCUMENTATION_INDEX.md. Tool count: ~195 base + ~830+ Pro (~1,025+ total; live count via WP_MCP_AI_Tool_Registry::get_tools() is authoritative). Provider count: 15 first-class language-model providers.

🆕 Previous Updates (v1.1.36 — July 2026)

June 29 – July 4, 2026 — EZuite Inventory Sync, Ralph Loop Orchestration, JetBooking/JetAppointment, Moonshot/Z.AI Parity, 45+ Bug Fixes 🏭🔄📅🧠🐛

• ✅ EZuite Inventory Sync Pro Toolkit. ERP-integrated inventory sync bridging EZuite ERP with WordPress/WooCommerce. Product pull, inventory query, item create/update, ERP settings, and CLI sync commands. Admin UI with connection selector, field mapping, and sync direction controls. Toolkit registered in Pro addon with auto-discovery.

• ✅ Ralph Loop CCT Migration & Orchestration. Circuit breaker pattern with configurable failure thresholds prevents cascading failures during cross-environment operations. Execution logger with step-by-step tracking and error reporting. CCT migration tools for safe JetEngine data transfers. Orchestration tools for coordinating multi-step workflows.

• ✅ JetBooking/JetAppointment Integration (8 Tools). Adapter layer for Crocoblock JetBooking and JetAppointment plugins. 8 new AI tools added to the Calendar Booking toolkit. 4 existing calendar tools enhanced with booking/appointment awareness. Full CRUD for appointments, bookings, and availability queries.

• ✅ Moonshot AI (Kimi) & Z.AI (GLM) Provider Parity. Both providers upgraded to full DeepSeek-level chat client capabilities — streaming, tool use, token tracking, and error handling. ZAI client service registered in DI container alongside client.baseten. Class loader updated for new provider adapters. WPCS formatting auto-fixed across all provider client files. Provider count now 15 first-class.

• ✅ Unified Sync Log Manager. Per-item audit trail across EZuite, FlowHub, and Shopify sync toolkits. Sync history with timestamps, status codes, and detailed error tracking. Status dashboards on toolkit settings pages. WP-CLI integration for log querying and export.

• ✅ Tool Presets Auto-Select & Chips Bar. Selected tools display as clickable chips below the tool selector with proper spacing. +N overflow toggle when too many chips are selected. Tool payload cap raised from 50 to 100 tools per assistant. Method-existence guard against tools lacking get_definition().

• ✅ HTTrack Cache Support & Place-to-Service Bridge. HTML place import tool now supports HTTrack cache directories for proper mirror parsing. URL backfill for mirrors without hts-cache directory. Place-to-Service bridge auto-creates bookable services during batch place import. Service sharing prevented when source URLs collide.

• ✅ FlowHub Per-Connection Overview. Remote Sites connection selector added to FlowHub config tab. Per-connection sync controls — run sync for individual connections instead of all-at-once. Proxy support via http_api_curl hook with connection-level proxy resolution.

• ✅ Generate Default Mapping + Read-Only Sync. One-click default mapping buttons for EZuite and FlowHub field mapping. Read-only sync direction option — pull data without pushing changes back to source.

• ✅ Web Search 429 Retry. Exponential backoff for rate-limited (HTTP 429) web search requests with configurable retry count and base delay.

• ✅ wp_mcp_ai_log Global Helper. Centralized logging function available globally, defined in includes/bootstrap/ for availability before class autoloading.

• ✅ JetEngine CCT Support for Remote WP Connections. Remote WordPress connections can now query and manage JetEngine CCT records — full CRUD support for remote CCTs via the remote_wp_connection tool.

• ✅ CCT Module API & CRUD Fixes (Cross-Toolkit). Replaced fragile jet_engine()->cct shorthand with canonical Module::instance() across EZuite, FlowHub, and Shopify CCT managers. Standardized CRUD using Factory/ItemHandler APIs. Added table_exists() fallback to is_cct_available() for lazy-init environments. Fixed get_cct_module false negatives when CCT module is active but not yet loaded.

• ✅ EZuite Sync Fixes. Missing return value with error diagnostics. API field mapping response key mismatch. Remote Sites connections not showing in toolkit settings (2 rounds). CCT auto-registration reading from wrong option key.

• ✅ FlowHub Sync Fixes. Silent success with 0 records now reports actual counts and errors. Proxy settings now correctly persist in Remote Sites connections. API auth header names corrected. Null CCT module guarded. Dry-run mode uses modules API instead of fragile shorthand. Location ID made optional with primary fallback.

• ✅ Shopify Sync Fixes. Catalog API-only connections blocked with clear error. CCT registration lifecycle uses explicit sanitize-update cycle. Meta fields populated in CCT registration. Column add_field routed through module->manager instead of raw jet_engine()->cct.

• ✅ CCT Infrastructure Fixes. Duplicate column errors prevented in ensure_columns across all three toolkits. add_field API replaced with direct SQL ALTER TABLE. Nonexistent get_item_by_slug() replaced with DB query. JetEngine CCT field definitions synchronized when columns missing.

• ✅ SQLite Meta Cache Explosion Fix. Unbounded meta cache growth on SQLite-backed sites resolved.

• ✅ masterminds/html5 Case Collision Fix. Class file collision on case-insensitive filesystems (Windows/macOS) resolved.

• ✅ Base-Version Guard Fix. Guard no longer incorrectly blocks toolkit admin pages in base+pro mode.

• ✅ HTTrack Import Robustness. URL resolution, redirect filtering, and type classification. File discovery bugs (SKIP_DOTS, isDot, substring skip-dir matching). Hex-encoded filename support. Non-GNU glob compatibility for subdirectory content. Improved mirror root detection.

• ✅ Necessity Gate & Safety Profile Fixes. Request context crash guarded against missing get_instance(). Safety profile trait autoload fixed — required from interface.

• ✅ Auto-Select Compute Fix. O(n*m) timeout on assistant edit page resolved.

• ✅ Tool Chip UI Fixes. PHP parse errors from unescaped double-quotes in inline JavaScript fixed.

• ✅ Documentation. Comprehensive abilities registration plan (~1,000 tools as WordPress Abilities). Laravel-scale deployment architecture proposal with Graphify ecosystem cross-references. WP.org submission prep (readme, index.php, canonical envelopes, READMEs). Agent context sweep — AGENTS.md, CLAUDE.md, .context/pro-vs-base.md updated. CHANGELOG v1.1.36 section.

• ✅ Housekeeping. Stale build artifacts and toolkit-addons directory cleaned.

• 📦 Versioning — bumped to 1.1.36 across mcp-ai-wpoos.php, WP_MCP_AI_VERSION constant (includes/bootstrap/constants.php), readme.txt, README.md, CHANGELOG.md, QUICK_REFERENCE.md, ROADMAP.md, and DOCUMENTATION_INDEX.md. Tool count: ~195 base + ~830+ Pro (~1,025+ total; live count via WP_MCP_AI_Tool_Registry::get_tools() is authoritative). Provider count: 15 first-class language-model providers (added Moonshot Kimi & Z.AI GLM at DeepSeek parity).

🆕 Previous Updates (v1.1.35 — June 2026)

June 27–29, 2026 — FlowHub + Shopify Sync Pro Toolkits, Necessity Gate Layer J, Local Voice STT, Remote Site Admin Blueprint, Bulk Import, Fixes 🏪🛒🛡️🎙️🌐

• ✅ FlowHub Inventory Sync Pro Toolkit (PR #5501). 6-tool cannabis dispensary management toolkit. FlowHub Products (flowhub_products) — query by location, category, brand, strain type. FlowHub Inventory (flowhub_inventory) — real-time levels, room/zone filtering, low-stock detection. FlowHub Locations (flowhub_locations) — address, contact, operational status. FlowHub Sync (flowhub_sync) — scheduled sync with webhook support. FlowHub Analytics (flowhub_analytics) — sales trends, top products, category performance. FlowHub Alert Manager — configurable low-stock and compliance alerts. FlowHub Settings (flowhub_settings) — API credentials, sync intervals, notifications. Connection resolver trait. Admin UI toggle on Features tab. Folder README at addons/pro/includes/tools/flowhub/README.md.

• ✅ FlowHub P1+P2 Enhancements (PR #5502). P1: Proxy support via http_api_curl hook for environments behind forward proxies. P2: CCT auto-registration — inventory data stored in JetEngine tables. Connection auth fixes: API header names corrected; class-name collision with WP_MCP_AI_FlowHub_Client resolved. API key decryption on credential update fixed (PR #5500). location_id made optional — defaults to primary location. Null guard against jet_engine()->cct for sites without JetEngine.

• ✅ Shopify Sync Pro Toolkit (PR #5502). 5-tool bi-directional e-commerce sync: Shopify Products (shopify_sync_products) — product sync with variant support. Shopify Orders (shopify_sync_orders) — order import with fulfillment tracking. Shopify Inventory (shopify_sync_inventory) — real-time stock sync. Shopify Analytics (shopify_sync_analytics) — cross-platform reporting. Shopify Settings (shopify_sync_settings) — API credential and sync config. Dashboard widget. Connection resolver trait. Admin UI toggle. Tool reference docs at docs/tools/shopify-sync-toolkit.md and docs/tools/flowhub-toolkit.md. Folder README at addons/pro/includes/tools/shopify-sync/README.md.

• ✅ Necessity Gate Layer J — Irreversibility-Weighted Safety Profiles. Pre-execution safety layer (WP_MCP_AI_Necessity_Gate) scores tool calls by irreversibility risk. Write operations (create/update/delete) assigned risk scores by resource type. Safety profile trait loaded from interface for clean autoload. Request context crash fix: guards against WP_MCP_AI_Request_Context lacking get_instance().

• ✅ Local Voice Embedded STT (PR #5498). Three pluggable browser-side speech-to-text backends: browser-native Web Speech API, Whisper.cpp (WASM), and Vosk (WASM). Offline-first — STT engines run entirely in the browser with no server dependency. Auto-detection selects the best available backend based on browser capabilities.

• ✅ Remote Site Administrator Blueprint. New 22-tool CRM-style blueprint remote-site-administrator.json in Site Creator toolkit. Pre-configured with remote_wp_connection, full JetEngine suite (11 tools), JetFormBuilder (2), site admin (4 — install_and_activate_plugin, install_and_activate_theme, update_option, create_post/update_post/delete_post), woo_products, generic_rest_api. Comprehensive system prompt with discovery→schema→execution workflow, tool reference, and 8 safety rules. Temperature 0.3, manage_options capability. Auto-discovered by Unified Blueprints page and import_site_creator_blueprint tool.

• ✅ Places & Calendar Bulk Import Tools (PR #5509). Batch import tools for Places and Calendar Booking toolkits. Places bulk import: batch geocode and store location records. Calendar bulk import: batch create events, appointments, bookings.

• ✅ CLI site-import Subcommand. wp mcp-ai site-import — multi-phase HTML mirror import for migrating static sites into WordPress. Handles page structure extraction, content mapping, and media sideloading.

• ✅ Voice Realtime Auto-Detect (PR #5508). WebRTC realtime voice auto-detection — automatically selects WebRTC or WebSocket transport. Duplicate message fix prevents echoed messages. VAD threshold improvements for better voice activity detection.

• ✅ Remote Connections Fixes (PR #5499). WordPress connection case handling normalized. FlowHub and Printful credential storage fixed — credentials correctly encrypted at rest. Printful connection type added to Remote Sites connection types.

• ✅ Token-Scoped Assistant Resolution (PR #5497). Token-scoped assistant now preferred over site default in resolver, ensuring bearer-token-authenticated requests use correct assistant config.

• ✅ User ID Empty Fallback (PR #5495). user_id resolution updated to use empty() instead of isset() for context fallback, fixing zero/empty-string bypass.

• ✅ Local Credential Token Mapping (PR #5493). Local credential token user ID mapping corrected for consistent identity resolution.

• ✅ Post Type Name Lengths (PR #5484). Post type names exceeding 20-char WordPress limit now truncated with validation warnings.

• ✅ OpenAI Image Deprecation Cleanup (PRs #5489-#5491). Removed deprecated DALL-E models. Applied chat model fallback to edit_image_via_responses_api. GPT-5.x chat models for image generation. Defaults and response_format handling fixed.

• ✅ Documentation. Agent context sweep — AGENTS.md (v1.7), CLAUDE.md (v2.7), .context/pro-vs-base.md updated. CHANGELOG v1.1.35 section.

• ✅ Housekeeping. Stale 1.1.34 build zips removed (6 files).

• 📦 Versioning — bumped to 1.1.35 across mcp-ai-wpoos.php, WP_MCP_AI_VERSION constant (includes/bootstrap/constants.php), readme.txt, README.md, CHANGELOG.md, QUICK_REFERENCE.md, ROADMAP.md, and DOCUMENTATION_INDEX.md. Tool count: ~195 base + ~810+ Pro (~1,005+ total; live count via WP_MCP_AI_Tool_Registry::get_tools() is authoritative). Provider count: 13 first-class language-model providers (unchanged).

🆕 Previous Updates (v1.1.33 — June 2026)

June 20–24, 2026 — WP 7.0 Connectors Credential Integration, nvoos-graphify v1.0.0, Security Fixes 🔌🧩🛡️🔒

• ✅ WP 7.0 Connectors Credential Integration (PR #5458). Credential_Resolver integrated into all 17 AI client get_api_key() methods (Anthropic, Baseten, DeepSeek, DigitalOcean, Gemini, Gemini Live, Google Maps, Hugging Face, Kimi, LM Studio, NVIDIA, OpenAI, OpenAI Realtime, OpenRouter, plus 3 service classes). Fallback chain: WP 7.0 Connectors → plugin settings → environment variables → PHP constants. New get_key_source() / get_key_source_label() methods added to Credential_Resolver. Credential source badges and WP 7.0 Connectors hints rendered in admin settings UI (Settings → Providers). All 13 provider API key field descriptions updated to mention alternative credential sources. Provider diagnostics now show a Key Source column indicating where each provider's API key originates. Settings health check counts credentials resolved via the Credential_Resolver. 17 Pro addon files updated to use Credential_Resolver::has_credentials() / get_api_key() for provider selection, OCR routing, vector store, embeddings, speech, and performance checks. All PHPCS-clean with zero new errors.

• ✅ nvoos-graphify v1.0.0 Release (PR #5456). Standalone nvoos-graphify plugin released at v1.0.0 — Plugin Check compliant and ready for distribution. nvoos-graphify-ai plugin released at v1.0.0-dev. Fixed 8 output-escaping errors in Section.php via printf() with esc_attr(). Fixed critical ->prepare() spread-operator bug in Db::listNodes() causing runtime fatals. Renamed vector column → embedding_vector to avoid MariaDB/MySQL reserved-word conflict; bumped DB version. Fixed snake_case→camelCase method calls across tools (GetNode, GetNeighbors, QueryGraph, GraphStats), controllers (SyncRemoteSource, REST), and cross-plugin integrations (nvoos-graphify-ai). Guarded missing Webhook driver and fixed Crypto namespace. Added message field to Enricher stub for detectable false-success states. Updated Remote README to reflect stub reality (no drivers, no SSRF, no encryption). Documented actual REST access model: read + guest token for reads, manage_options for export/write.

• ✅ Security Dependencies (PR #5457). guzzlehttp/guzzle 7.10.0 → 7.12.1 (CVE-2026-55568, CVE-2026-55767). guzzlehttp/psr7 2.11.0 → 2.12.1 (CVE-2026-55766). guzzlehttp/promises 2.3.0 → 2.5.0. undici npm override tightened from >=7.28.0 to >=8.5.0 to match resolved version and prevent downgrades.

• ✅ npm Security (PR #5438). 29 npm security alerts resolved across 14 packages. undici: TLS bypass (CVE-2026-9697) and cache info disclosure (CVE-2026-9678) — override applied to 11 addon lock files. http-proxy-middleware: CRLF injection (CVE-2026-55603) — override >=3.0.7. nodemailer: raw option bypass (GHSA-p6gq-j5cr-w38f) — bumped ^8.0.9 → ^9.0.1. webpack-dev-server: HMR interception (CVE-2026-9595) — override >=5.2.5. dompurify: ALLOWED_ATTR pollution (GHSA-cmwh-pvxp-8882) — bumped ^3.4.9 → ^3.4.11. Fixed critical duplicate overrides key in root package.json that silently discarded all 39 security overrides.

• ✅ Bug Fix — WP All Import/Export Pro Tools (commit 0de3cdf). Four require_once paths in the Pro bootstrap map (addons/pro/mcp-ai-wpoos-pro.php) were missing -pro- in the filenames, referencing class-wp-mcp-ai-tool-*.php instead of the actual class-wp-mcp-ai-pro-tool-*.php files on disk, causing fatal errors when those tools loaded.

• ✅ Bug Fix — Tool Status Label Loader (commit 749ffce). Replaced fragile @file_get_contents() with explicit set_error_handler('__return_true') / restore_error_handler() in both copies of load_tool_status_labels() (includes/class-wp-mcp-ai-tool-registry.php and includes/admin/sections/class-wp-mcp-ai-section-tools.php). The @ operator does not reliably suppress warnings across all PHP 8 configurations (e.g. xdebug.scream, custom error handlers), and leaked warnings corrupt MCP JSON-RPC HTTP responses.

• ✅ Dependency Updates. 15 Dependabot version bumps across the monorepo: Composer (guzzlehttp/psr7 2.11.0→2.12.1). npm/Pro (stripe 14.25.0→22.2.3, csv-parse 5.6.0→7.0.0, p-queue 8.1.1→9.3.0, @puppeteer/browsers upgrade). npm/SaaS Controller (@wordpress/eslint-plugin 25.2.0→25.4.1, zod 3.x→4.4.3, @tanstack/react-query upgrade, wrangler upgrade). npm/Cloud Worker (stripe upgrade, vitest 2.x→4.1.9, @cloudflare/workers-types upgrade). npm/Docs Hub (@typescript-eslint 8.x→8.62.0, @types/node→26.0.0). GitHub Actions (codecov/codecov-action 4→7, softprops/action-gh-release 2→3).

• ✅ Housekeeping. Stale 1.1.31 build zips removed (PR #5437). Stale 1.1.32 build zips removed (6 files). SPA addon ZIPs rebuilt with updated security overrides and dependency bumps. nvoos-graphify standalone ZIP built at v1.0.0. nvoos-graphify-ai ZIP built at v1.0.0-dev.

• 📦 Versioning — bumped to 1.1.33 across mcp-ai-wpoos.php, WP_MCP_AI_VERSION constant (includes/bootstrap/constants.php), readme.txt, README.md, and CHANGELOG.md. Provider count: 13 first-class language-model providers (unchanged). Tool count: ~195 base + ~795 Pro (~990 total; live count via WP_MCP_AI_Tool_Registry::get_tools() is authoritative).

• ✅ MCP Initialize: Assistant-Scoped Instructions & Model Preferences (PR #5462). Both the base /mcp-ai/v1/mcp endpoint and all 29 Pro toolkit MCP servers now carry the assistant's system prompt, professional role context, model preferences (modelPreferences — supported by Zed, Claude Desktop, Cursor), and knowledge base references in the initialize handshake instead of a generic WordPress site description. serverInfo.name becomes the assistant's display title when an assistant_id is provided. New filter wp_mcp_ai_mcp_initialize_instructions for integrator customization. Zero breaking changes — behavior is identical when no assistant_id is passed. Unix Theory P0-compatible: canonical return envelope and two-gate sanitisation rule unaffected (protocol-layer change only).

June 17–19, 2026 — Content Format Templates, Result Delivery Pipeline, Featured Images, Provider Timeout Fixes 📝📬⏱️🔧

• ✅ Content Format Templates & Featured Image Generation (PR #5433). Content Format Template CPT (mcp_ai_content_format_template) with user-editable blog post templates. New WP_MCP_AI_Content_Template_Engine generates Anthropic-optimised XML-sectioned AI prompts. New WP_MCP_AI_Featured_Image_Service with 3-provider fallback (OpenAI DALL-E → Google Gemini → Cloudflare AI) and 5 image styles. Image Generation Provider dropdown on the CPT metabox. Provider settings (model, quality, aspect ratio) now respected instead of hardcoding dall-e-3/hd. AI prompts updated across all schedule presets to instruct image generation before post creation. 5 workflow presets updated with image gen nodes, node reorder, and featured_image_id wiring. Template resolution via Content Template Engine in dispatch_assistant_run. resolve_node_template_variables for cross-node variable passthrough. 5 default templates seeded on activation.

• ✅ Result Delivery Pipeline (PR #5425). New WP_MCP_AI_Result_Delivery_Service (1,056 lines) routes successful schedule results to 8 delivery channels: email, Slack, Discord, Telegram, SMS, Paper Store, WordPress post, webhook. Per-channel formatting and sending with deliver_success() and deliver_failure() pathways. result_delivery schema added to schedule CRUD. Result Delivery section in schedule edit modal UI. Delivery status appended to run history and result envelopes. Pre-configured Paper Store delivery for 8 research/audit presets. Both success and failure paths now deliver — previously only failures were surfaced.

• ✅ ECA Document Generation (PR #5423). ECA Consolidate & Add page with document generation tools. PHPCS alignment fixes across ECA admin pages.

• ✅ Duplicate Posts & Provider Image Fixes (PR #5434). WordPress delivery channel removed from weekly_blog_post_writer and weekly_blog_topic_research presets to prevent duplicate draft posts when the AI already calls create_post during assistant runs. Systemic guard in Result Delivery Service skips WordPress delivery when AI tool calls include create_post or save_post. Provider image settings now properly flow through Content Template Engine and Featured Image Service.

• ✅ 6 Provider Clients Now Respect Timeout (PR #5431). DeepSeek, Baseten, DigitalOcean, OpenRouter, Kimi, and Cloudflare clients had hardcoded 60-second timeouts ignoring the admin Request Timeout (seconds) setting. All six resolve_timeout() methods now read WP_MCP_AI_Admin_Settings, matching the existing Anthropic/Gemini/Ollama pattern. Kimi retains its provider-specific kimi_timeout override.

• ✅ Schedule Trigger Stability (PRs #5429, #5430). Trigger crash on rest_do_request() wrapped in try/catch with pre-flight REST server check (PR #5429). User context restored and filters cleaned up in all error paths. display and result_delivery fields added to ajax_get_schedules response for edit modal persistence. sanitize_result_delivery preserves channels wrapper so all consumers read saved delivery config (PR #5430). Chat-channel channel field preserved through sanitize_delivery_channels. ajax_trigger_schedule wrapped in ob_start + try/catch to capture PHP warnings that corrupt JSON. Trigger debug output logged to browser console when WP_DEBUG is on.

• ✅ Paper Store Delete Confirmation Fix (PR #5432). Hidden inputs added to record delete confirmation form so handle_delete_record() receives required POST fields. handle_delete_collection() updated to read from $_REQUEST for GET-based admin-post.php links.

• ✅ ECA Settings Menu & Attachment Fix (PR #5421). ECA settings menu placement corrected. Attachment upload crash in ECA admin pages resolved.

• ✅ npm CI & Jest Resilience (PR #5428). @babel/plugin-transform-modules-systemjs override pinned to v7 (v8 is ESM-only, breaking eslint on Node 18/20). Root package-lock.json regenerated with override applied. jest.config.js probes for setup files at load time with graceful fallback to empty array. jest.setup.js at repo root. || exit 0 on test:coverage for CI. npm ci lockfile sync for chat-spa, media-studio, docs-hub, comic-reader, and saas-controller.

• ✅ Dependency Updates. 14 safe Dependabot version bumps across docs-hub, cloud-worker, saas-controller, Pro (react 19.2.6→19.2.7, vitest 4.1.5→4.1.9, stripe, workers-types, eslint 8.59.2→8.61.1, php-stubs 6.9.1→7.0.0, mailparser 3.7.1→3.9.9, validator 13.12.0→13.15.35). 8 npm audit CVEs fixed: nodemailer, tar, tar-fs. Security overrides added to 8 addon package.json files. phpoffice/phpspreadsheet lock synced 5.7.0→5.8.0.

• 📦 Versioning — bumped to 1.1.32 across mcp-ai-wpoos.php, WP_MCP_AI_VERSION constant (includes/bootstrap/constants.php), readme.txt, README.md, and CHANGELOG.md. Provider count: 13 first-class language-model providers (unchanged). Tool count: ~195 base + ~795 Pro (~990 total; live count via WP_MCP_AI_Tool_Registry::get_tools() is authoritative).

June 16–17, 2026 — Media Command Center, Pro SPA v2, Workflow Presets, npm CVEs, Gemini 3.1 Flash 🎬💬🧰🔒🎨

• ✅ Media Command Center (PR #5402). Top-level "NV Media" admin menu with command center for managing media workflows. Template and preset management. PHPCS cleanup and implode() WP_Error crash fix in the templates tab.

• ✅ Pro SPA v2 Migration & Rich Rendering (PRs #5401, #5412, #5414). Pro SPA v2 ported to feature parity with chat-spa and old pro-spa. Rich markdown rendering for assistant responses. Per-assistant scoping dropdown and agent selector. Conversations set as primary view; threads marked read-only after archival. CSS class-name mismatches fixed, causing previously unstyled UI to render correctly. Cache-bust version bump and admin layout fit within WordPress chrome (PR #5416). Bumped to v2.0.1.

• ✅ 34 Missing Workflow Preset Tools (PR #5418). All missing workflow presets implemented across 10 toolkits (AI Tool Builder, Analytics, Architect Agent, Architectural Design, Calendar Booking, CRM, Document Generation, DJ Management, Social Media, Video Production). Orphaned media tools created and Upwork tool availability fixed. Full PHPCS compliance and comprehensive test suite validating all 36 new tools (class existence, interface implementation, slug resolution, metadata, capability flags).

• ✅ npm Audit CVE Fixes (PR #5419). 12 CVEs resolved across the dependency tree: vite (CVE-2026-53571 server.fs.deny bypass), launch-editor (CVE-2026-53632 NTLMv2 disclosure), markdown-it (CVE-2026-48988 smartquotes DoS), ws (memory exhaustion), js-yaml (quadratic DoS), form-data (CRLF injection), hono (path traversal, CORS, Set-Cookie), dompurify (XSS sanitization bypasses), @babel/core (arbitrary file read), @opentelemetry/core (unbounded memory), joi (RangeError DoS). ajv bump reverted to restore ESLint 8 compatibility. Root npm audit: 51→0 vulnerabilities.

• ✅ Gemini 3.1 Flash Image Default (PR #5404). Default Gemini image generation model upgraded to gemini-3.1-flash-image.

• ✅ Media Toolkit Blueprints & Scheduler Presets (PRs #5398, #5411). Blueprints and scheduler presets added to the media toolkit. Media Toolkit sync integrated into the Data Management page. Private get_media_presets() visibility fixed to prevent fatal error when called externally.

• ✅ OpenAI/DeepSeek stream_options (PR #5400). stream_options payload added to OpenAI and DeepSeek streaming API calls, enabling proper usage-tracking inclusion during streaming responses.

• ✅ Tool Result Cost Calculation (PR #5395). Tool result costs now computed from token counts when not explicitly provided by the provider, ensuring accurate cost attribution in agentic loops.

• ✅ Agentic-Loop Cost Tracking (PR #5394). Missing provider pricing entries added across the cost-tracking layer, fixing under-reporting in multi-turn agentic workflows.

• ✅ Vite CVE Fixes (PR #5403). vite pinned to ^8.0.16 in devDeps to resolve CVE-2026-53571 (server.fs.deny bypass) and CVE-2026-53632 (NTLMv2 disclosure via launch-editor middleware). Pro SPA v2 production assets built and dist output un-gitignored.

• ✅ 1,658 PHPCS Lint Fixes (PR #5397). Full lint pass across base plugin and Pro addon (44 toolkits). Short ternaries converted to full, count() moved out of loops, docblock capitalisation normalized, require_once paths corrected for renamed optimization and tool files.

• ✅ CI Disk Space Cleanup (PR #5417). Free disk space step added to all build workflows to prevent runner disk exhaustion during large ZIP assembly.

• ✅ Data Integrity Fixes. PII pseudonymisation logic restored after being lost during a merge conflict resolution (PR #5398). CPT data store post type mismatch for CRM entities resolved (PR #5396).

June 12–15, 2026 — Chat SPA Phase 8, PM Toolkit A–D, CRM Duplicates & Hygiene, DietPi Toolkit, LibreChat Addon, Context Windows, WP 7.0 Bridge 🪲💬📊🔧🛡️

• ✅ Chat SPA Phase 8 — Message Actions & Content Enrichment (PRs #5381, #5383, #5390). Conversations sidebar with assistant scoping in the Pro Chat SPA. Auto-create thread when posting to a non-existent thread. Message action cards on every assistant bubble: edit, delete, regenerate, copy, and content enrichment. Pro SPA chat routed through chat-client endpoint; threads marked read-only after archival.

• ✅ Project Management Toolkit Enhancement Phase A–D (PR #5370). Shared WP_MCP_AI_PM_Engine powering the new Command Center dashboard with real-time task visibility, status filters, and bulk actions. Work Ingestion panel on the tasks tab for importing external tasks. 28 new AI tools spanning task management, resource allocation, timeline generation, and reporting.

• ✅ CRM Toolkit — Duplicates, Email Hygiene & Analytics (PRs #5362, #5367, #5368). Duplicate detection engine with safe one-click merge and bulk merge from the new Duplicates tab. Email hygiene module: classify domains by reputation, manage exclusion/priority lists, and auto-prune. Top Customers and Top Clients analytics tools with dedicated Command Center tabs. Merged status filter and column on leads admin list. Inline tag editing and email priority/exclude actions. Repair CRM data tool fixing broken dates, generic titles, and impersonal sender names. Refresh All Sources button on Support tab. LinkedIn & Upwork external sourcing with LinkedIn as a Remote Sites OAuth 2.0 connection type.

• ✅ DietPi Pro Toolkit Phases 0–3 (PRs #5346, #5348, #5350). Foundations: CPT/CCT scaffolding, admin settings page, 19 core tools for system monitoring, package management, and service control. Phase 2: backup, update, storage, and dashboard tools. Phase 3: provisioning automation, infrastructure blueprints, and SSH proxy. Registered as an MCP server with Feature subtab toggle.

• ✅ LibreChat Addon (PR #5336). New addons/librechat/ addon with code interpreter, speech-to-text/text-to-speech services, and web search reranker. Added to SPA build workflow and addon ZIP build pipeline.

• ✅ Layer I Guardrails — Stay-on-Target Jailbreak Prevention (PRs #5340, #5344). Harness Layer I intercepts prompt-injection and role-change attacks. Activated per-assistant via the LLM Harness metabox.

• ✅ Context Window Management — All 13 Providers (PRs #5335, #5348, #5352). Pre-flight context-window validation added to every AI provider client (OpenAI, Anthropic, Gemini, DeepSeek, OpenRouter, Baseten, Kimi, DigitalOcean, NVIDIA, Cloudflare, Hugging Face, LM Studio, Ollama). Shared validate_context_window() helper in the language model router. tiktoken integration for accurate token counting with estimator metabox on the Assistant editor. Token-budget tool capping — tools exceeding remaining budget are auto-excluded. Chat parity drift detection workflow. Model limits synced with June 2026 canonical catalog.

• ✅ WP 7.0 Connectors API Bridge (PR #5387). Bridge for provider credentials enabling forward-compatibility with the WordPress 7.0 Connectors API. Classes load unconditionally to prevent fatal errors on WP < 7.0.

• ✅ Chat Transcript & Agent Memory Retention (PRs #5356, #5357). WP_MCP_AI_Transcript_Retention — configurable retention policies and automatic cleanup cron. WP_MCP_AI_Memory_Retention — agent memory lifecycle management with pruning and retention windows.

• ✅ Pro Toolkit Optimizations Phase 1–3 (PRs #5355, #5356, #5357). Phase 1: Chat Channels and Social Media toolkits. Phase 2: Healthcare, Ecommerce, Calendar, and Orchestration. Phase 3: Document Generation and QMS. All classes wired into their toolkit init files with 767-line PHPUnit test suite.

• ✅ OAuth & API Connection Management (PR #5351). Disconnect buttons for OAuth and API connections. Auto-clear OAuth tokens when provider credentials change on save.

• ✅ SPA Reliability Sweep (PRs #5371–#5379). Register thread REST endpoints (fixes createThread crash). Fix SPA router context error. Fix createThread event serialization crash. Register POST /threads/{id}/messages endpoint. Pass model and profile to createThread. Fix SPA bootstrap tools, turn count, and REST route. Fix SPA enqueue paths and PM blueprints. Restore Zed-inspired SPA after accidental revert. Build SPA bundle and fix import paths.

• ✅ Core & Provider Fixes. 12 test suites fixed and SPL autoloader ordering corrected (#5392). Toolkit MCP tools returning empty inputSchema for image tools (#5391). OOS engine fatal error from undefined sendRequest (#5389). Agentic loop tool result persistence across all providers (#5354). OpenAI real-time SSE streaming (#5327). Bridge credential resolver on WP < 7.0 (#5388). Stale provider validation lists rejecting DeepSeek (#5323). Schedule preset data mismatches (#5329).

• ✅ CRM & Data Fixes. CRM configured sources count showing 0 for valid Gmail connections (#5386). Slider/range fields not rendering (#5361). 3 CPT slugs exceeding 20-character limit (#5361). Support ticket CPT rename mcp_ai_support_ticket → mcp_ai_ticket (#5360). Memory tools capability flags (#5364). Extended Cognition review issues (#5363). Playbook orphans and sync timeouts (#5322, #5325).

• ✅ Addon Fixes. Docs Hub: async rebuild stalls, browse critical error, broken Intuit privacy link (#5359). Cloudways Dashboard: removed Requires Plugins header (#5353); base-active check in _is_ready() (#5385). Pro Integrations subtab overwriting redirect (#5347). Performance monitor: buttons, reports, test execution restored (#5366). DietPi Toolkit PHPCS and PHPCompat fixes (#5346).

• ✅ Security. guzzlehttp/psr7 pinned to ^2.10.2 for CVE-2026-49214 (#5352). esbuild pinned to ^0.28.1; integrity check vulnerability in Pro addon (#5382). shell-quote bumped to >=1.8.4 for CVE-2026-9277 (#5330). vitest/vite/esbuild alerts in schedule-anything-spa (#5338).

• ✅ Infrastructure. redirect_canonical no longer breaks well-known endpoints (#5349). tool-status.txt reads guarded against missing files (#5348). WPCS lint in section-tools and tool-registry (#5348). ESLint no-var in performance-admin.js (#5366). PHPCS in CRM analytics (#5320). Chat parity workflow YAML and permissions fix (#5336).

• ✅ Blueprint & Doc Updates. 46 healthcare-style blueprints converted to CRM-style format (#5373). All legacy flat-format blueprints upgraded to canonical (#5372). v1.1.29 comprehensive documentation refresh across all surfaces (#5365). 66 proposal statuses audited (#5335). README updated with PM toolkit and repository map (#5370). 75 PHPCS violations auto-fixed in Pro bootstrap (#5321).

• 📦 Versioning — bumped to 1.1.31 across mcp-ai-wpoos.php, WP_MCP_AI_VERSION constant (includes/bootstrap/constants.php), readme.txt, README.md, and CHANGELOG.md. Provider count: 13 first-class language-model providers (unchanged). Tool count: ~195 base + ~795 Pro (~990 total; live count via WP_MCP_AI_Tool_Registry::get_tools() is authoritative).

🆕 Previous Updates (v1.1.32 — June 2026)

June 7–11, 2026 — Bug-Fix & Stabilisation Sweep 🪲🔧⚡

• ✅ Chat Bubble Assistant Dropdown (PR #5333). Fixed settings UX: chat_bubble_assistant_id field on the Chat Bubble settings page (wp-admin/admin.php?page=wp-mcp-ai-dashboard&tab=general&subtab=chat_bubble) changed from a plain <input type="number"> to a proper <select> dropdown populated with all published assistants. Added get_assistant_options() helper method to WP_MCP_AI_Section_Chat_Client matching the pattern used by default_assistant in General → Core Settings. Users can now select assistants by name instead of manually typing numeric IDs.

• ✅ Context-Window Pre-Flight Validation — All 13 Providers (PR #5328). Added pre-flight context-window validation to all AI provider clients: OpenAI, Anthropic, Gemini, DeepSeek, OpenRouter, Baseten, Kimi, DigitalOcean, NVIDIA NIM, Cloudflare, Hugging Face, LM Studio, and Ollama. Shared validate_context_window() helper method in WP_MCP_AI_AI_Client_Base. Integrated tiktoken for accurate token counting with estimator metabox on the Assistant editor. Token-budget tool capping prevents exceeding model limits. Context-window management documentation at docs/developer/architecture/context-window-management.md.

• ✅ OpenAI SSE Streaming Fix (PR #5327). Fixed stream_options payload flag that prevented OpenAI real-time SSE streaming from triggering. The include_usage flag was incorrectly nested, causing the OpenAI API to ignore the streaming request and return a non-streamed response.

• ✅ Schedule Preset Data Mismatches (PR #5329). Fixed schedule preset data mismatches where presets would lose configuration data after save. Improved error logging for preset operations with structured log contexts.

• ✅ Playbook Orphan Cleanup & Batching (PRs #5322, #5325). Fixed orphan playbook accumulation where deleted parent records left unreachable children. Added JetEngine validation before deletion to prevent data corruption. Fixed sync timeouts for large playbook sets by implementing batch processing with configurable chunk sizes. playbook_delete_batch_timeout now uses iterative deletion instead of single-query operations.

• ✅ Stale Provider Validation Lists (PR #5323). Fixed provider validation lists in multiple locations that rejected DeepSeek and newer providers. Updated WP_MCP_AI_Section_General::validate() and WP_MCP_AI_REST_Chat_Controller provider checks to use dynamic provider discovery via WP_MCP_AI_Admin_Settings::get_available_providers() instead of hardcoded arrays. Also fixed default provider validation in CRM settings and Pro REST controllers.

• ✅ CRM Activity Titles, Due Dates & Block API v3 (PR #5320). Fixed CRM activity post titles not displaying correctly when created via AI tools. Fixed due date calculations for recurring activities. Migrated CRM admin blocks to WordPress Block API v3 (apiVersion: 3) for compatibility with WP 6.9+.

• ✅ OpenAI-Compatible Client — DeepSeek Parity (PR #5315). Enhanced all OpenAI-compatible chat clients (OpenRouter, Baseten, Kimi, DigitalOcean, NVIDIA NIM) to parity with the DeepSeek client: added max_completion_tokens support, temperature normalisation, top_p handling, and proper stop sequence forwarding.

• ✅ Voice & Embedded LLM Missing Assets (PR #5319). Added missing .min.js asset files for voice recording/transcription scripts and embedded LLM worker scripts. These files were referenced by wp_register_script() but missing from the build output, causing 404 errors on sites using minified assets.

• ✅ Chat Config messagesEndpoint & Tool Count Guard (PR #5318). Fixed chat config messagesEndpoint pointing to the wrong REST route in certain configurations. Added a tool count guard that returns a clear error message when an assistant has more tools configured than the provider's limit, preventing silent failures.

• ✅ Chat Debug Console Fixes (PRs #5316, #5317). Fixed chatDebugMode using loose equality (== instead of ===) so PHP's string '1' is accepted alongside JavaScript's boolean true. Fixed legacy chat debug console not displaying when enabled via admin settings — the debug panel container was hidden by a CSS rule that only targeted the new SPA debug view.

• ✅ OOS Bridge, Embedding Fatal & SSE Headers (PR #5313). Fixed OOS bridge initialization failing when the core framework was loaded before WordPress user context was available. Fixed embedding service fatal error when API key was not configured — now returns a graceful WP_Error. Fixed SSE header warnings in PHP 8.1+ caused by header_remove() being called after output started.

• ✅ Memory Cookie-Check Nonce Fix (PR #5312). Fixed "Cookie check failed" error on the admin Test Assistant memory drawer. The memory REST controller was generating a nonce on init (before the user session was available), producing an invalid nonce for authenticated users. Nonce generation moved to the wp hook.

• ✅ Chat Transcript Tests — 4% to 87% Pass Rate (PR #5310). Fixed chat transcript REST controller tests. Root causes: test factories not creating posts with the correct post_type, missing WP_REST_Server initialization, session key normalisation mismatches, and permission callback assertions testing the wrong user role. Pass rate improved from ~4% (3/80) to 87% (70/80).

• ✅ Graphify Related Content Leak (PR #5291). Fixed graphify related content leaking to wrong page sections — content isolation tightened so graph nodes only render within their designated container element.

• ✅ Model Limits — June 2026 Canonical Catalog (PR #5331). Synced model limits (max tokens, max output tokens, rate limits) with the June 2026 canonical catalog across all providers.

• ✅ Security — shell-quote CVE-2026-9277 (PR #5330). Bumped shell-quote to >=1.8.4 via npm overrides to fix CVE-2026-9277 (command injection via insufficient escaping).

• 📦 Versioning — bumped to 1.1.29 across mcp-ai-wpoos.php, WP_MCP_AI_VERSION constant (includes/bootstrap/constants.php), package.json, readme.txt Stable tag, README.md, and docs/DOCUMENTATION_INDEX.md. Provider count: 14 first-class language-model providers (OpenAI, Gemini, Anthropic, DeepSeek, OpenRouter, Baseten, Kimi, DigitalOcean, NVIDIA NIM, Cloudflare, Hugging Face, LM Studio, Ollama, Flowhub). Tool count: ~195 base + ~795 Pro (~990 total; live count via WP_MCP_AI_Tool_Registry::get_tools() is authoritative).

📋 Previous Releases

For full details on all releases, see CHANGELOG.md.

🚀 Features

Note: Some features require third-party plugins (WooCommerce, JetEngine, Elementor, etc.). See 🔌 What You Lose Without Third-Party Plugins for details.

Assistant & conversation tools

• 🧠 Create AI Assistants via a custom post type (mcp_ai_assistant)

• 👔 Professional & Team Templates - Deploy assistants from ~311 pre-built profession templates spanning 12 industry categories, or create entire teams of specialists with one click. Includes backend testing for professions, teams, and assistants before public deployment.

• 🚀 Getting Started Wizard - Guided 4-step onboarding (/wp-admin/admin.php?page=wp-mcp-ai-getting-started) that walks new users through provider setup and use-case selection. Selecting a preset (Content Creator, Customer Support, E-commerce, SEO & Research, Developer Copilot, Media & Creative Studio, Site Administrator, or General Purpose) seeds a fully-configured assistant with tools, system prompt, and tuned temperature — ready to use immediately.【F:includes/admin/class-wp-mcp-ai-onboarding-wizard.php†L1-L53】【F:assets/js/onboarding-wizard.js†L1-L303】

• 🔄 Automatic synchronization to JetEngine Custom Content Types when available (CPT → CCT)

• 💬 Chat interface via [mcp_ai_chat assistant="ID"]

• 🧰 Per-assistant defaults for model, temperature, system prompt, and knowledge attachments with permission-aware download URLs

• ⚡ Build reusable prompt shortcuts with optional tool targeting and inline descriptions so operators can trigger common tasks with one click.【F:includes/assistants/class-wp-mcp-ai-assistant-cpt.php†L893-L1048】【F:includes/class-wp-mcp-ai-shortcode.php†L430-L693】【F:assets/js/chat.js†L600-L666】

• 🧊 Elementor widgets for embedding chat surfaces, onboarding content, and MCP dashboards inside Elementor

Language routing & knowledge management

• 🔁 Route conversations through OpenAI or Gemini using a provider-aware language model router

• 🎯 Enhanced Gemini API integration: list models dynamically, count tokens for budget management, create embeddings for RAG/semantic search, and streaming support for real-time responses【F:docs/reference/api/gemini/gemini-api-enhancements.md†L1-L100】

• 🧠 Assistant knowledge base management with Media Library files and optional vector store IDs

• 📚 OKF (Open Knowledge Format v0.1) engine with 6 MCP tools for curated, deterministic knowledge with cross-link navigation — complementary to vector/RAG stores

• 🔎 Perform lightweight web searches (DuckDuckGo or Brave) without leaving the assistant conversation

• 🌐 Crawl4AI job runner tool for large-scale content gathering workflows

Media generation & transcription

• 🔊 Generate speech audio via OpenAI's Text-to-Speech API and save the result to the Media Library

• 🎵 Generate instrumental music using Google Gemini Lyria with controls for genre, mood, tempo, and instrumentation

• 🎨 Generate on-brand imagery with OpenAI's Images API, honouring the configured response format (including GPT-Image-1's url responses) and storing the files as WordPress attachments

• 🖼️ Generate images with Cloudflare Workers AI using Stable Diffusion, Flux-2 Dev, Leonardo AI (Lucid Origin, Phoenix 1.0), and other text-to-image models with configurable dimensions and generation parameters

• 🖼️ Vectorize raster images (PNG, JPEG, WebP, GIF) to SVG format using @neplex/vectorizer with configurable quality settings - perfect for logos and icons

• 🎨 Comprehensive graphic editing with Graphic Editor Plus combining local operations (logo overlay, smart resize) and AI-powered features (style transfer, background removal, enhancement)

• 🏗️ Pro: Generate professional architectural drawings (floor plans, elevations, sections) with building codes, dimensions, and material specifications - designed for construction professionals

• 🎧 Transcribe or translate uploaded audio with OpenAI's speech-to-text endpoints

Commerce & finance workflows

• 🛍 WooCommerce-aware tools (fetch orders or products, requires WooCommerce)

• 📊 Finance-ready QuickBooks Online reporting tool for surfacing Profit and Loss, Balance Sheet, and other statements inside assistant conversations【F:includes/tools/class-wp-mcp-ai-tool-get-quickbooks-report.php†L15-L214】【F:includes/admin/class-wp-mcp-ai-admin-settings.php†L906-L955】

• 🖥️ Pro: QuickBooks Desktop sync via QODBC relay API — connect to QuickBooks Desktop through a Windows relay server for data synchronization

• 🛒 Pro: Shopify integration with auto-resolved connections — connection_id auto-resolved from assistant context, covering products, orders, customers, inventory, and catalog tools

• 🚗 Pro: Vehicle estimation tools — VIN decode (NHTSA vPIC), image-to-repair-estimate pipeline, and car wash package pricing engine (always available)

• 📸 Pro: Listing image download tools — bulk-download Google Maps, Facebook, and Instagram business listing images into the Media Library or ZIP

Slash Commands & Workflow Automation ⭐ NEW

• ⚡ 8 Core Commands: /help, /next-task, /ship, /clean-content, /optimize-perf, /sync-docs, /workflow - Command-line style interface for content management

• 🔄 Workflow Orchestrator: Multi-step workflow execution with state management, conditional logic, and human-in-the-loop checkpoints

• 🛠️ 21 Pro Toolkit Commands: Specialized commands for E-commerce (6), Social Media (6), and Video Production (6) toolkits

• 🎯 7 Automated Workflows: Pre-built workflow templates for abandoned cart recovery, social media campaigns, video marketing, inventory management, and more

• 🔐 Security: Capability-based authorization, rate limiting, comprehensive audit logging

• 💡 Integration: JavaScript autocomplete, REST API endpoint, WP-CLI support

• Documentation → | Pro Commands →

Chat Channels & Messaging Integration ⭐ NEW

• 💬 Chat Channels Toolkit (47 Tools): Integrate with 11 platforms - Telegram, WhatsApp, Slack, Discord, Microsoft Teams, Facebook Messenger, Apple Messages for Business, Google Chat/Spaces, Twitter/X, Office 365 (Outlook + OneDrive), iCloud Drive

• 📧 Office 365 Integration ⭐ NEW: Send and retrieve Outlook mail, list/download/upload OneDrive files via Microsoft Graph API (5 tools)

• ☁️ iCloud Drive Integration ⭐ NEW: List, download, and upload iCloud Drive files via a configurable gateway service (3 tools)

• 🌐 Unified Broadcasting: Send messages across multiple platforms simultaneously with unified_channel_broadcast tool

• 🏠 WebChat Rooms: Custom post type for real-time collaborative chat rooms with AI assistant assignment

• 📝 Message Persistence: JetEngine CCT integration for permanent message history

• 🔊 WebRTC Support: Self-hosted WebRTC signaling via WordPress REST API for voice/video

• 🤖 AI-Powered Rooms: Assign dedicated assistants to chat rooms for automated support

• Chat Channels Guide → | WebChat Setup →

Communications & outreach

• ✉️ Mailjet-powered outbound email automation with granular capability enforcement and sender defaults configurable in the MCP settings.【F:includes/tools/class-wp-mcp-ai-tool-send-mailjet-email.php†L19-L405】【F:includes/admin/class-wp-mcp-ai-admin-settings.php†L1008-L1054】

• 📅 Google Workspace automations for creating calendar events and searching connected Gmail inboxes directly from assistant workflows.【F:includes/tools/class-wp-mcp-ai-tool-create-google-calendar-event.php†L1-L200】【F:includes/tools/class-wp-mcp-ai-tool-search-gmail.php†L1-L200】

• 🧾 JetFormBuilder orchestration for listing forms, reviewing submissions, and proxying REST calls on behalf of assistants (requires JetFormBuilder)

• 📚 JetEngine REST route reference tool for surfacing endpoint metadata inside AI workflows

• 🧱 Ready for extension with ChatKit integration

Integrations, security & controls

• 🔧 Tool Registry for registering PHP functions callable by the AI

• ⚙️ JetEngine integration for dynamic content queries (requires JetEngine)

• 🔌 JetEngine 3.8 MCP Server Bridge ⭐ NEW - JSON-RPC 2.0 client bridges NV oOS into JetEngine's native MCP Server with 7 new Pro tools for CPT/taxonomy/meta field creation, relations management, site context grounding, and prompt template access. MCP-first dispatch with REST v2 fallback.

• 🤝 Agent-to-Agent (A2A) Protocol ⭐ NEW - Full A2A protocol making NV oOS assistants discoverable and interoperable with any A2A-compliant agent. /.well-known/agent.json discovery, JSON-RPC 2.0 server with task state machine, A2A client for remote agent delegation, push notification webhooks.

• 📊 Agent Command Center ⭐ NEW - Unified agent management dashboard with 7 tabs: Overview (KPI cards, live status), Activity Log, Active Tasks, Approvals (human-in-the-loop), Analytics (Chart.js with real per-agent metrics), Uptime & Health, and Strategy (efficiency scoring with recommendations).

• 💬 Floating Chat Bubble ⭐ NEW - Configurable floating chat bubble widget for Elementor and Gutenberg. 4 position variants, 3 sizes, bounce/pulse animations, dark mode, WCAG focus states, sessionStorage persistence.

• 🧷 Granular control over allowed attachment MIME types for chat uploads

• 🔐 Secure REST API endpoints

• 🔑 Root Security Key - Optional wp-config.php constant that can be enabled during emergency shutdown to require authentication before re-initializing the plugin. Provides an additional layer of protection against unauthorized reactivation after security incidents.【F:docs/features/security/root-security-key.md†L1-L511】【F:includes/class-wp-mcp-ai-root-security-key.php†L1-L360】

• 🛰 Assistant directory endpoint that advertises MCP tool/resource capabilities and negotiates Server-Sent Events handshakes for clients such as LM Studio or Claude Desktop.【F:includes/class-wp-mcp-ai-rest.php†L520-L666】【F:includes/class-wp-mcp-ai-rest.php†L1690-L1772】

• 📝 Full JSON-RPC 2.0 MCP endpoint (/mcp) for standards-compliant remote client communication

• 🔑 Configurable API credentials and defaults for OpenAI, Gemini, and Anthropic (with subscription tier support for Team/Enterprise plans and custom base URLs)

• 🤖 ChatGPT’s connector beta currently requires an Auth0 tenant; the plugin’s assistant credentials are compatible with LM Studio, Claude, and other MCP clients that support bearer headers directly.【F:docs/reference/api/mcp-server-authentication.md†L22-L46】

• 🌐 Mesh networking for distributed compute pooling across multiple WordPress sites. Server-to-server architecture enables anonymous and authenticated users to benefit from shared AI resources, budget pooling, and workload distribution across 100+ trusted peer sites. Backend assistants coordinate mesh operations via secure inter-site keys while maintaining user attribution and audit trails for compliance.【F:docs/features/federation/mesh-compute-pooling.md†L1-L615】【F:includes/tools/class-wp-mcp-ai-tool-query-remote-site.php†L1-L237】

• 🔗 Federation & Discovery - Decentralized AI capability network allowing WordPress sites to publish their capabilities via well-known endpoints (/.well-known/ai-peer) and discover peer sites through directory services. Supports peer registration, health verification, search & ranking by capability/region/policy, and automatic cron-based health monitoring. Enable federation to join the network or run your own directory service for private peer discovery.【F:docs/features/federation/federation-discovery.md†L1-L511】【F:FEDERATION-IMPLEMENTATION-SUMMARY.md†L1-L381】

• 🧾 Optional logging of chat interactions, tool executions, and API errors

• 🧮 Built-in per-user usage tracking for provider/model billing summaries

• 🧩 Developer hooks and filters for integrating custom behaviours

• ⏱ Per-site request timeout control with sensible minimum enforcement

• 🗑 Toggleable uninstall cleanup to purge stored assistants and settings automatically

Performance & reliability

• ⚡ Client-side message bundling (800ms window) to reduce API calls and server load【F:docs/user-guides/chat/message-bundling-feature.md†L1-L80】

• 🎯 Intelligent token overflow handling with automatic model switching (gpt-4.1-mini → Gemini 2.0 Flash)【F:docs/features/tools/presets/high-token-tool-handling.md†L1-L80】

• 📡 Server-Sent Events (SSE) support for real-time streaming responses and job notifications【F:docs/features/streaming/ENABLE-SSE-STREAMING.md†L1-L100】

• 🌊 Real-time job status updates via SSE streaming and webhook notifications for async operations【F:docs/features/async-jobs/job-notification-system.md†L1-L100】

• 🔧 Symfony Process Component - Modern process execution framework replacing direct exec() calls in Pro addon tools for enhanced security, timeout management, and error handling【F:includes/services/class-wp-mcp-ai-process-service.php†L1-L220】【F:docs/history/2025/implementations/symfony-phases/SYMFONY_PHASE2B_PROCESS_INTEGRATION.md†L1-L100】

• 🔄 Server-side WP-Cron polling for long-running tasks (Crawl4AI, background jobs)

• 💾 Chat history persistence with localStorage (24h) and optional JetEngine CCT storage【F:docs/user-guides/chat/chat-history-persistence.md†L1-L50】

• ⚙️ Optimized settings page with external CSS stylesheet (240 lines added to admin-settings.css) and request-level caching for improved admin performance【F:assets/css/admin-settings.css†L1-L984】【F:includes/admin/class-wp-mcp-ai-admin-settings.php†L27-L32】

Settings Management ⭐ NEW

• 🔧 Robust Settings System - 7-step save process with automatic backups, validation, and cache management ensures settings persist correctly across all tabs and subtabs【F:includes/admin/class-wp-mcp-ai-settings-dashboard.php†L262-L410】

• 🔍 Health Check - Run 6 diagnostic checks to verify settings integrity, provider configuration, and system status with GOOD/WARNING/CRITICAL status indicators【F:includes/admin/sections/class-wp-mcp-ai-section-advanced.php†L1500-L1650】

• 💾 Export Settings - Download all plugin settings as timestamped JSON files for backup or migration to other sites【F:docs/admin-guides/settings-management.md†L40-L80】

• 📤 Import Settings - Upload and validate settings from previously exported backups with automatic pre-import backup and 5-step validation【F:docs/admin-guides/settings-management.md†L85-L135】

• 🗑️ Clear Cache - One-click clearing of static cache, object cache, and transients when settings changes don't take effect【F:docs/admin-guides/settings-management.md†L140-L165】

• ↩️ Reset to Defaults - Safely reset all settings to default values with automatic backup before reset【F:docs/admin-guides/settings-management.md†L170-L200】

• 🔒 Security - File size validation (max 5MB), MIME type checking, JSON validation, and comprehensive input sanitization【F:includes/admin/class-wp-mcp-ai-settings-dashboard.php†L970-L1030】

• 📊 Automatic Backups - Every save operation creates a timestamped backup (keeps last 5) for emergency recovery【F:includes/admin/class-wp-mcp-ai-settings-dashboard.php†L285-L295】

• 🛡️ Data Protection - 3-layer protection (section filtering, merge strategy, sensitive key filtering) prevents accidental data loss when saving from tabs/subtabs【F:docs/admin-guides/settings-management.md†L230-L280】

• 📖 Pro Toolkits - Enable and configure 8 specialized Pro toolkits (650+ tools) including Project Management, Document Generation, Health & Wellness, CRE Debt & Securitization, and more【F:docs/admin-guides/pro-settings-toolkits.md†L1-L650】

➡️ Complete Documentation: Settings Management Guide | Quick Reference | Visual UI Guide | Pro Toolkits Guide

🧠 Memory & Tool Stack Overview

Model defaults

Global settings capture the default provider, model, and timeout used when assistants are created, ensuring every conversation inherits stable generation behaviour until explicitly overridden. These defaults ship with sensible values for OpenAI and Gemini out of the box and can be tailored from the NV oOS settings screen.【F:includes/admin/class-wp-mcp-ai-admin-settings.php†L36-L77】

Base knowledge

Each assistant can preload Media Library files and optionally link to an external vector store, giving the model persistent project context before a chat begins. Editors manage these knowledge sources from the assistant post type via the “Base Knowledge” meta box, which supports multiple attachments and vector store identifiers.【F:includes/assistants/class-wp-mcp-ai-assistant-cpt.php†L892-L1002】

Available tools

The tool registry boots with a curated catalogue of content, commerce, automation, and research utilities, then exposes hooks so developers can register their own providers. During initialisation the registry loads each bundled tool class—ranging from JetEngine accessors to Crawl4AI jobs and Mailjet automations—and makes them callable within conversations.【F:includes/class-wp-mcp-ai-tool-registry.php†L74-L220】

Chat-client Memory Drawer

The chat front-end exposes a persistent Memory Drawer (assets/js/chat-memory-drawer.js) with three tabs:

• Memories — browse, pin, and delete stored context items; 🧠 badge auto-appears on any assistant message that used a memory tool.

• Scope — set the active wing/room scope for subsequent memory operations in the current session.

• Audit — lazy-loaded audit trail from WP_MCP_AI_REST_Chat_Memory_Controller::audit().

The drawer is wired to the REST proxy at /mcp-ai/v1/chat-memory/ and receives real-time updates via the memory_event SSE frame emitted by the agentic loop. Pagehide auto-capture stores the session state before tab close. Two gates control access: site-wide filter wp_mcp_ai_chat_memory_enabled and per-user meta wp_mcp_ai_chat_memory_enabled. Full reference: docs/features/memory/chat-client-integration.md.

Retroactive Transcript Mining

WP_MCP_AI_Transcript_Mining_Job retrospectively extracts memories from past chat transcripts. Enqueue a job via POST /mcp-ai/v1/transcript-mining/jobs (admin-only), poll progress with GET /jobs/{id}, or cancel with POST /jobs/{id}/cancel. Full reference: docs/features/memory/transcript-mining.md.

LLM Harnessing Subsystem

Seven opt-in per-request layers (includes/harness/) improve response quality without changing existing tool behaviour. Activated per-assistant via the LLM Harness metabox. Layers: A Prompt/Cue → B Reasoning Trace → C Tool Routing → D Retrieval → E Self-Refine → F Memory Scoping + PII Filter → G Eval Scheduler. Pro Layer H exports fine-tune curricula as OpenAI JSONL. Full reference: docs/features/llm-harness.md.

Workflow families & tool combos

The core plugin ships with a centrally registered tool catalogue that lets assistants mix and match capabilities into cohesive workflows without additional coding. Teams can chain authoring, media, research, commerce, marketing, and operational tools to deliver end-to-end outcomes inside a single conversation.

• Content & knowledge production – Combine submit_document_prompt, search_content, and search_attachments to gather source material, then follow up with save_post, create_wpcode_snippet, or get_rankmath_seo for structured drafting and optimisation.

• Media generation & transcription – Pair generate_openai_image, generate_gemini_image, vectorize_image, or graphic_editor_plus with generate_openai_speech and transcribe_openai_audio to build multimedia assets that flow into editorial or marketing outputs. Use vectorize_image to convert logos to scalable vectors, and graphic_editor_plus for comprehensive image editing with both local and AI-powered operations.

• Research & situational awareness – Chain discovery helpers like web_search, run_crawl4ai_job, reliefweb_reports, get_gdacs_events, and get_nhc_active_storms to assemble briefing packs before drafting follow-up actions.

• Commerce & finance operations – Use WooCommerce and finance tools such as create_woo_product, get_woo_products, get_woo_recent_orders, crawl4ai_price_lookup, get_import_duty, and quickbooks_report to coordinate merchandising, pricing, and bookkeeping reviews.

• Marketing & analytics insights – Combine measurement tools including google_analytics_report, get_google_business_insights, get_facebook_instagram_insights, get_linkedin_insights, and get_tiktok_insights to guide campaigns and reporting.

• Publishing & outreach automations – Trigger distribution via post_facebook_instagram, post_google_business_update, post_linkedin_update, post_tiktok_video, send_group_email, send_mailjet_email, send_telegram_message, send_whatsapp_message, and schedule_notify_sms once plans are ready.

• Integrations & scheduling – Connect external systems with create_google_calendar_event, search_gmail, list_jetengine_rest_routes, invoke_jetengine_route, and run_openai_external_action as part of larger automations.

• Operations & diagnostics – Close the loop with create_cron_job, list_cron_jobs, get_cron_job, delete_cron_job, check_wp_cli, purge_cache, purge_cloudflare_cache, purge_varnish_cache, get_site_summary, get_site_health, get_system_logs, get_update_status, and OpenAI usage/log review helpers for monitoring and maintenance.

• Automation & scheduling workflows – Agents can autonomously schedule background tasks with create_cron_job, monitor scheduled operations via list_cron_jobs and get_cron_job, and clean up outdated automations with delete_cron_job. Combine with cache management tools (purge_cache, purge_cloudflare_cache, purge_varnish_cache) to orchestrate content publishing workflows where agents schedule posts, then automatically invalidate caches at publication time.

🛠 Built-in tools & automations

The assistant registry ships with a comprehensive catalogue of editorial, marketing, commerce, and operational helpers. The tables below outline every bundled tool and the slug assistants call when orchestrating workflows.

Content & knowledge workflows

Media generation & transcription

Research & situational awareness

Commerce & finance operations

Marketing & analytics insights

Publishing & outreach

Integrations & scheduling

Operations & diagnostics

What the Cron Manager means to AI agents

The Cron Management Suite transforms AI assistants from reactive responders into proactive automation orchestrators. By providing full control over WordPress's background task scheduler, agents can:

Autonomous Task Scheduling

• Schedule content publishing workflows to go live at optimal times without human intervention

• Automate recurring maintenance tasks like cache clearing, database optimization, or backup operations

• Coordinate multi-step operations that span hours or days by chaining scheduled hooks

Intelligent Monitoring & Self-Management

• List and inspect all scheduled tasks to understand what automation is currently active

• Audit who created each task and when it's scheduled to run next

• Identify and remove outdated or redundant scheduled tasks to maintain system health

Real-World Agent Workflows

1. Content Calendar Automation - An agent helping with content strategy can schedule posts to publish at researched optimal engagement times, set up recurring social media cross-posts, and schedule follow-up email campaigns.

2. Site Maintenance Orchestration - When troubleshooting performance issues, agents can schedule off-peak cache purges, coordinate database cleanup tasks, and set up recurring health check notifications.

3. Business Process Automation - Agents can schedule recurring report generation, periodic data syncs with external systems, and automated backup verification checks.

Technical Implementation The cron manager tracks all scheduled tasks in wp_mcp_ai_cron_jobs option with full audit trails including:

• Job ID for unique identification

• Hook name and sanitized arguments

• Schedule type (single-run or recurring interval)

• Creation timestamp and user attribution

• Next execution time for monitoring

Jobs are automatically pruned when they complete (single-run) or are manually removed (recurring), keeping the tracking database clean. All cron operations require manage_options capability, ensuring only authorized users can delegate automation authority to agents.【F:includes/class-wp-mcp-ai-cron-manager.php†L12-L280】

Each tool inherits the assistant context and authenticated user from the REST layer, making it easy to layer custom permissions or extend behaviour via the documented filters and actions.【F:includes/class-wp-mcp-ai-rest.php†L236-L360】【F:includes/class-wp-mcp-ai-rest.php†L1124-L1198】

Need per-tool prerequisites or capability callouts? Consult docs/reference/tools/tool-reference.md for a detailed matrix of every bundled integration.

Tool Status Labels

The Tools Manager page displays status labels beside tool names to indicate their development stage and stability:

Status labels are displayed as 3-letter abbreviations (e.g., "STA" for stable, "BET" for beta) to keep the UI compact.

Important: Tools marked with the bug status are automatically disabled when the plugin loads. This prevents problematic tools from being used until issues are resolved. Administrators can manually re-enable them from the Tools Manager if needed for testing.

Status labels are managed via the tool-status.txt file in the repository. To assign a status label to a tool:

1. Open docs/tool-status.txt in a text editor

2. Add a line in the format: tool_slug = status_label

3. Save the file - changes appear immediately in the Tools Manager

Example:

create_post = stable
web_search = beta
generate_openai_image_validated = experimental
problematic_tool = bug

This file-based approach allows quick status updates without code changes, making it easy for maintainers to reflect tool maturity as development progresses. The automatic disabling of buggy tools provides an additional safety layer to prevent issues in production environments.

🗨️ Front-end chat surfaces

NV oOS ships multiple ways to embed assistants on the front end:

• Classic chat shortcode – [mcp_ai_chat] renders the bundled interface with attachment uploads, tool invocation feedback, and optional guest access via allow_guests="true". When guest mode is enabled, the shortcode provisions a temporary token and injects it into the JavaScript bootstrap so visitors without WordPress accounts can continue chatting while still respecting capability checks and attachment safety limits.【F:includes/class-wp-mcp-ai-shortcode.php†L132-L258】【F:includes/class-wp-mcp-ai-shortcode.php†L188-L226】

• Floating chat bubble ⭐ NEW – A configurable floating button that sits at a screen corner and opens a chat panel powered by the [mcp_ai_chat] shortcode. Available as both an Elementor widget and a Gutenberg block. Supports 4 position variants, 3 sizes, auto-open delay, session persistence, dark mode, and WCAG keyboard navigation.【F:includes/elementor/class-wp-mcp-ai-elementor-chat-bubble-widget.php†L1-L200】【F:includes/blocks/chat-bubble/block.json†L1-L50】

• Elementor widgets – Drop the chat UI anywhere Elementor is active, pair it with intro/FAQ blocks, and surface dashboard telemetry without custom code. The chat widget mirrors the shortcode controls (including allow_guests), and companion widgets expose onboarding content, usage timers, provider quick links, and activity feeds for operational views.【F:includes/elementor/class-wp-mcp-ai-elementor-widget.php†L79-L138】【F:includes/class-wp-mcp-ai-elementor-integration.php†L48-L98】【F:includes/elementor/class-wp-mcp-ai-elementor-chat-intro-widget.php†L47-L140】【F:includes/elementor/class-wp-mcp-ai-elementor-chat-usage-timer-widget.php†L48-L226】【F:includes/elementor/class-wp-mcp-ai-elementor-dashboard-activity-feed-widget.php†L48-L167】

Guest tokens are honoured by the REST endpoints through the X-WP-MCP-AI-Guest header or guest_token parameter, allowing the chat shortcode and Elementor widget to make authenticated requests on behalf of public visitors without exposing persistent credentials.【F:includes/class-wp-mcp-ai-rest.php†L289-L307】【F:includes/class-wp-mcp-ai-rest.php†L2088-L2104】

Chat History Persistence

The chat interface automatically persists conversation history to the browser's localStorage, preventing data loss when users navigate away or refresh the page. Conversations are:

• Automatically saved after each user message and assistant response

• Automatically restored when returning to the chat page (within 24 hours)

• Stored per assistant so different assistant conversations remain separate

• Server-side storage available with JetEngine - See note below about optional JetEngine integration

Server-Side Chat Transcript Storage (Requires JetEngine)

⚠️ Third-Party Plugin Required: JetEngine (not included with NV oOS)

Without JetEngine, chat conversations are only stored in browser localStorage (client-side, 24-hour retention). To enable permanent server-side chat transcript archiving:

1. Install and activate the JetEngine plugin (third-party, paid plugin from Crocoblock)

2. Enable the Custom Content Types module in JetEngine settings

3. NV oOS will automatically provision the ai_chat_transcripts CCT for permanent storage

What you get with JetEngine:

• ✅ Permanent server-side chat transcript storage

• ✅ Cross-device conversation access

• ✅ Admin visibility into chat history

• ✅ Database-backed chat logs for compliance/auditing

Without JetEngine:

• ⚠️ Chat history only stored in browser localStorage

• ⚠️ Limited to 24-hour retention

• ⚠️ No cross-device synchronization

• ⚠️ Lost if browser data is cleared

See docs/user-guides/chat/chat-history-persistence.md for complete details on the persistence mechanism, data structure, and troubleshooting.

📦 Installation

ℹ️ Plugin Directory Status
This plugin is currently pending approval in the WordPress Plugin Directory. We are committed to maintaining high quality and security standards throughout the review process. You can install the plugin manually from our GitHub repository or wait for the official WordPress Plugin Directory listing.

🚀 Getting Started Wizard
After activating the plugin, you'll be redirected to a 4-step setup wizard that walks you through connecting an AI provider, choosing a use case, and creating your first assistant — all in under 2 minutes. The wizard creates fully-configured assistants with tools, system prompts, and tuned temperatures so your site is working out of the box. You can access the wizard any time at NV oOS → Getting Started or directly at /wp-admin/admin.php?page=wp-mcp-ai-getting-started.

🌱 Beginner 3-Step Install (Try it on Your PC)

No live site. No API costs. No risk. The fastest way to try NV oOS is on your own computer using a free local WordPress environment. Follow the full walkthrough on the NV Digital Solutions blog: How to Test NV oOS on Your Own PC Using Local + Downloading the Plugin →

1. Install a local WordPress environment — Local by WP Engine or WordPress Studio is the easiest option (one-click install, no server config). Alternatives: XAMPP, MAMP, or DevKinsta.

2. Download the NV oOS plugin zip — grab the latest release from GitHub Releases (look for mcp-ai-wpoos-x.x.x.zip), or use the Code → Download ZIP button for the current development snapshot.

3. Upload, activate, and run the wizard — in your local WordPress dashboard go to Plugins → Add New → Upload Plugin, select the zip, activate it, and follow the 🚀 Getting Started Wizard. For a free local AI model (no API key needed), install LM Studio and point the wizard at http://localhost:1234.

🗺 Single-file auto-installer (roadmap) — A single cross-platform installer that bootstraps the entire stack automatically is on the roadmap. See the App / Plugin Distribution Proposal for current status and the plan.

Requirements

Minimum Requirements:

• WordPress 6.0+

• PHP 7.4+ (PHP 8.0+ recommended)

• MySQL 5.7+ or MariaDB 10.3+

Optional Requirements for Enhanced Features:

• Node.js 14+: Required for image vectorization tools (vectorize_image tool)

• PHP Functions: proc_open, proc_close, proc_terminate (for Node.js integration and Process Service)

  • These functions are often disabled on shared hosting for security

  • Can be enabled on Cloudways via Application Settings (see troubleshooting guide)

• JetEngine Plugin: For CCT storage and advanced content management tools

• WooCommerce: For e-commerce integration tools

• Elementor: For visual page builder widgets

Note: The plugin works without optional requirements, but some features will be disabled. See deployment troubleshooting for enabling disabled PHP functions.

For Developers (GitHub Clone)

✅ Production-Ready Repository
This repository includes production-optimized vendor dependencies with classmap-authoritative autoloading configured by default in composer.json. You can clone and activate immediately without running composer. The composer install command is only needed if you want to update dependencies or add development tools.

⚡ Use a shallow clone
A full clone of this repository is ~10 GB due to its long history. Use --depth 1 to download only the latest snapshot (~500 MB) — much faster and smaller. If you later need the full history, run git fetch --unshallow.

git clone --depth 1 https://github.com/nvdigitalsolutions/mcp-ai-wpoos.git

If you're cloning from GitHub:

Option 1: Cloudways and Managed Hosting (Recommended)

For Cloudways and similar managed hosting platforms, clone directly into the WordPress plugins directory:

# SSH into your server
# Navigate to WordPress plugins directory
cd /home/master/applications/YOURAPP/public_html/wp-content/plugins/

# Clone the repository (production-ready, no composer needed!)
# Use --depth 1 for a fast shallow clone (recommended for production)
git clone --depth 1 https://github.com/nvdigitalsolutions/mcp-ai-wpoos.git
cd mcp-ai-wpoos

# Verify you're in the correct directory
pwd  # Should show the plugins path

# Optional: Only needed for frontend asset rebuilding or development
# npm install && npm run build

# Optional: Only run if you need to update dependencies or add dev tools
# Note: Autoloader optimization is now configured by default in composer.json
# composer install --no-dev

⚠️ Cloudways Important Notes:

• Always clone directly into /home/master/applications/YOURAPP/public_html/wp-content/plugins/

• Do NOT clone elsewhere and then move/copy - this causes getcwd() failed errors

• Replace YOURAPP with your actual Cloudways application name

Option 2: Local Development or VPS

For local development or standard VPS hosting:

# Option A: Clone directly into WordPress plugins directory (recommended, production-ready!)
cd /path/to/wordpress/wp-content/plugins/
# Use --depth 1 for a fast shallow clone (recommended for production)
git clone --depth 1 https://github.com/nvdigitalsolutions/mcp-ai-wpoos.git
cd mcp-ai-wpoos
# Ready to activate! No composer or npm needed for production use.

# Option B: Clone and copy (also production-ready!)
# Use --depth 1 for a fast shallow clone (recommended for production)
git clone --depth 1 https://github.com/nvdigitalsolutions/mcp-ai-wpoos.git
cd mcp-ai-wpoos
cp -r . /path/to/wordpress/wp-content/plugins/mcp-ai-wpoos/

For Development Only:

# Only if you need to rebuild assets or modify dependencies:
npm install && npm run build
# Note: Autoloader optimization is now configured by default in composer.json
composer install --no-dev

Optional: Strip Dev Files for Production

If you are deploying via git clone to a production server with anti-malware / EDR scanning, the working tree will contain test fixtures that embed verbatim attack-payload literals (XSS canaries, SQL-injection samples, prompt-injection strings) used by the security test suite. These can occasionally trip signature-based scanners.

For a clean production tree, run the bundled strip script after cloning:

# Preview what would be removed
bin/strip-dev-files.sh --dry-run

# Remove tests/, docs/, bin/, .github/, .bmad/, .context/, examples/,
# phpunit.xml.dist, phpcs.xml.dist, dev configs, etc.
bin/strip-dev-files.sh

The script mirrors the exclusion list in .distignore (used for the WordPress.org SVN deploy) and the export-ignore rules in .gitattributes (used for GitHub-distributed ZIPs). It is idempotent and refuses to run on a working tree with uncommitted changes (override with --force).

Note: Do not run this on a development checkout — it removes the test suite, docs, and build tooling. It is intended for deploy targets that only run the plugin.

Final Steps

1. Activate Open Operator System Complete (NV oOS) from WordPress admin

2. You now have the complete version with all ~1,566 tools (~303 base + ~1,263 Pro; live count via WP_MCP_AI_Tool_Registry::get_tools() is authoritative)

What you get from the repository clone:

• ✅ The full codebase — all ~1,566 built-in tools ready to use (~303 base + ~1,263 Pro; live count via WP_MCP_AI_Tool_Registry::get_tools() is authoritative)

• ✅ Single plugin activation (not separate base + pro)

• ✅ Pro features automatically available (no separate Pro plugin to install)

Notes:

• The repository includes mcp-ai-wpoos-base.php and addons/pro/mcp-ai-wpoos-pro.php which are used for building separate distributions but do NOT appear as separate plugins when cloning

• Only the main plugin file (mcp-ai-wpoos.php) has a plugin header in the repository

• The build script adds headers to the other files when creating standalone distributions

Standard Installation

1. Upload mcp-ai-wpoos.zip to /wp-content/plugins/

2. Activate NV oOS from the WordPress admin

3. Go to Settings → NV oOS

4. Enter your OpenAI API key

5. Create a new “AI Assistant” in AI Assistants

6. Add [mcp_ai_chat assistant="123"] to a page or post

Optional: JetEngine Integration

⚠️ Third-Party Plugin (Not Included): JetEngine is a paid plugin from Crocoblock

JetEngine is completely optional - NV oOS works perfectly without it. However, if you want server-side chat transcript storage:

1. Purchase and install JetEngine separately

2. Enable the Custom Content Types module in JetEngine settings

3. NV oOS will automatically provision the ai_chat_transcripts CCT for permanent chat storage

What works WITHOUT JetEngine:

• ✅ All core AI assistant features

• ✅ Chat interface and conversations

• ✅ ~300 base tools (more with optional third-party plugins)

• ✅ MCP server functionality (/wp-json/mcp-ai/v1/)

• ✅ Browser-based chat history (localStorage, 24 hours)

• ✅ OpenAI/Gemini/Anthropic/Ollama/Hugging Face/Cloudflare integrations

What requires JetEngine:

• ❌ Server-side chat transcript storage (chat history only in browser without it)

• ❌ 5 JetEngine-specific tools (see 🔌 Optional Tools & Dependencies)

🔌 What You Lose Without Third-Party Plugins

NV oOS works perfectly with vanilla WordPress, but certain features require third-party plugins (sold separately). Here's exactly what you lose without each plugin:

Without JetEngine (Crocoblock - Paid Plugin)

Lost Features:

• ❌ AI metaboxes for JetEngine CPTs/Taxonomies - No AI assistant integration on JetEngine edit screens

• ❌ Research & Add pages - No AI-powered content creation with automatic field mapping

• ❌ Server-side chat transcript storage - Chat history only stored in browser localStorage (24 hours)

• ❌ Cross-device chat synchronization - No database-backed conversation history

• ❌ Admin chat history access - Cannot view/audit conversations from admin panel

• ❌ Assistant CCT synchronization - Assistants only in WordPress CPT (MCP server still works perfectly)

Lost Tools (5 tools):

• get_jetengine_items - Query JetEngine custom post types

• list_jetengine_rest_routes - List JetEngine REST API routes

• invoke_jetengine_route - Execute JetEngine REST operations

• get_jetformbuilder_forms - List JetFormBuilder forms (also requires JetFormBuilder)

• get_jetformbuilder_submissions - Get form submissions (also requires JetFormBuilder)

✅ Still Works: All core features, MCP server, ~300 base tools, AI conversations

Get JetEngine →

Without WooCommerce (Free Plugin)

Lost Features:

• ❌ E-commerce automation - Cannot create or manage products via AI

• ❌ Order management - Cannot query or analyze orders

• ❌ Product catalog access - Cannot search or update product data

Lost Tools (3 tools):

• create_woo_product - Build draft WooCommerce products with AI-generated descriptions, pricing, and images

• get_woo_products - Search and retrieve product catalog with pricing and stock status

• get_woo_recent_orders - Summarize recent orders with billing details and totals

Use Cases Lost: E-commerce content generation, order fulfillment assistance, product merchandising

Get WooCommerce →

Without Elementor (Freemium Plugin)

Lost Features:

• ❌ Template management - Cannot list or reference Elementor templates via AI

• ❌ Elementor widgets - Cannot use pre-built chat/dashboard widgets (shortcodes still work)

Lost Tools (2 tools):

• get_elementor_templates - List Elementor library templates with status, type, and edit links

• import_elementor_template_kit - Import Elementor template kits

Lost UI Components:

• Elementor Chat Widget

• Elementor Chat Intro Widget

• Elementor Dashboard Widgets (Tool Matrix, User Capabilities, Activity Feed, etc.)

✅ Still Works: Standard [mcp_ai_chat] shortcode, all AI features

Get Elementor →

Without Rank Math SEO (Freemium Plugin)

Lost Features:

• ❌ SEO analysis - Cannot query SEO scores or optimization recommendations

• ❌ Schema data access - Cannot retrieve structured data for posts

Lost Tools (1 tool):

• get_rankmath_seo - Get SEO scores, focus keywords, robots metadata, and schema details for posts

Use Cases Lost: AI-powered SEO content optimization, SEO audit assistance

Get Rank Math →

Without WPCode (Freemium Plugin)

Lost Features:

• ❌ Code snippet management - Cannot create or update code snippets via AI

• ❌ Custom functionality automation - Cannot automate adding hooks, filters, or custom code

Lost Tools (1 tool):

• create_wpcode_snippet - Create or update code snippets with validation and activation control

Use Cases Lost: AI-assisted custom development, automated code snippet generation

Get WPCode →

Without Simple JWT Login (Free Plugin)

Lost Features:

• ❌ JWT token generation - Cannot generate JWT bearer tokens for headless WordPress integrations

Lost Tools (1 tool):

• generate_simple_jwt_token - Generate JWT bearer tokens for authenticated API access

Use Cases Lost: Headless WordPress authentication, mobile app integration, SPA authentication

Get Simple JWT Login →

Summary: Third-Party Plugin Dependencies

Total Impact: Without these plugins, you lose 13 tools but retain ~300 base tools and all essential AI assistant functionality.

Base Version (Default)

NV oOS runs in Base Version mode by default, providing ~300 essential tools that work with vanilla WordPress without requiring any third-party plugins:

Base Version includes ~300 essential tools that work with vanilla WordPress:

• Content management (search, save posts, attachments)

• AI media generation (images via OpenAI/Gemini, speech, transcription, video)

• Research tools (web search, weather, disaster alerts)

• Site operations (health checks, logs, cron jobs, cache management)

• WordPress-native email (via wp_mail)

• Image manipulation (resize, crop, rotate, convert, vectorize to SVG)

• Graphic editing (local operations and AI-powered transformations)

• Profession and assistant management

• GitHub integration tools

• Google Maps Platform tools

Base Version excludes 31 tools requiring third-party plugins or external APIs:

• Third-party WordPress plugins (13 tools) - See 🔌 What You Lose Without Third-Party Plugins for details

  • WooCommerce tools (3)

  • JetEngine/JetFormBuilder tools (5)

  • Elementor tools (2)

  • RankMath/WPCode/Simple JWT Login tools (3)

• External API services (18 tools) - Require API credentials

  • Google services (5)

  • Social media integrations (8)

  • External messaging services (4)

  • QuickBooks (1)

Full Version Installation (Opt-in)

To enable the Full Version with all third-party integrations and external API tools, add this constant to your wp-config.php file:

define( 'WP_MCP_AI_BASE_VERSION', false );

📖 See BASE-VERSION.md for the complete tool list and customization options.

When to use Base Version:

• Starting fresh with WordPress

• Testing or development environments

• Simpler installations without external dependencies

• Sites that don't need e-commerce or advanced integrations

• Don't want to purchase/install third-party plugins

When to use Full Version:

• Production sites with WooCommerce, JetEngine, or Elementor already installed

• Sites needing social media automation (requires API credentials)

• Advanced workflows requiring external APIs

• Need server-side chat transcript storage (requires JetEngine)

📖 See detailed breakdown: 🔌 What You Lose Without Third-Party Plugins

📚 Documentation

NV oOS includes comprehensive documentation covering all aspects of the plugin. Documentation reorganized June 2026 — Unix-theory separation of concerns. All docs sorted into 12 purpose-driven directories with zero content loss.

📖 Documentation Hub

• Documentation Hub ⭐ Start here - Central navigation with organized categories

• Documentation Index - Complete map of all 1,600+ documentation files

• Architecture Overview - System architecture (15 providers, ~1,566 tool classes, 36 REST controllers)

• Request Flow Walkthrough - End-to-end chat request lifecycle trace

• Quick Reference Guide - Fast access to common tasks and commands

Essential References

• Tool Reference - All ~1,566 tools documented (~303 base + ~1,263 Pro; live count via WP_MCP_AI_Tool_Registry::get_tools() is authoritative)

• REST API Documentation - Complete API reference with examples

• Testing & Quality Report - Test results and code quality analysis

📦 Archive

• Historical Documentation — 50+ archived files from 2024-2025 development

• Docs Archive - Consolidated implementation history and superseded documentation

For New Users

• 🚀 Getting Started Wizard ⭐ NEW — 4-step guided setup that connects your AI provider, selects a use case, and creates a ready-to-use assistant in under 2 minutes. 8 presets available: Content Creator, Customer Support, E-commerce, SEO & Research, Developer Copilot, Media & Creative Studio, Site Administrator, General Purpose.

• Use Cases & Quickstart Guides ⭐ NEW - Comprehensive guide covering 7 major use cases with step-by-step quickstarts

• 5-Minute Quick Start - Get started immediately: from zero to first chat

• Setup Checklist - Step-by-step installation and configuration

• Remote Client Quickstart - Connect Claude Desktop, LM Studio, or other MCP clients

• Best Practices - Recommended usage patterns and optimization tips

For Developers

• Testing & Quality Report - Test suite results (2,106 tests, 73.4% pass rate), code quality analysis, security audit

• Code Review Master - Comprehensive code quality analysis (95/100 score)

• Action Items - Prioritized development tasks (180+ hours)

• Authentication Guide - Authentication methods and security

• MCP JSON-RPC 2.0 Endpoint - Model Context Protocol implementation

For Administrators

• Deployment Troubleshooting - Common issues and solutions

• Multisite Support - WordPress multisite configuration

• Rate Limit Protection - API rate limiting setup

• Mesh Routing Guide - Intelligent compute routing across sites and providers

• Federation & Discovery - Decentralized AI capability network with peer discovery and well-known endpoints

Performance & Optimization

• Message Bundling - Client-side message optimization

• High Token Tool Handling - Agentic loop token management

• Job Notification System - Real-time async job updates

• Chat Performance Optimizations - Complete performance guide

• Mesh Routing Guide - Intelligent compute routing across sites and providers

Historical Documentation

• Archive Directory - 95+ historical documents organized by category:

  • implementations/ - Implementation summaries and technical details

  • phases/ - Development phase documents

  • fixes/ - Bug fix summaries and issue resolutions

  • features/ - Feature documentation

  • code-reviews/ - Code review reports

  • testing/ - Test infrastructure documentation

⚙️ Configuration Checklist (Action Items)

Complete these after installation to unlock every integration point:

• Add your OpenAI API key in Settings → NV oOS → OpenAI API Key so API calls are authorised.

• Add your Gemini API key in Settings → NV oOS → Gemini API Key if you plan to route assistants through Gemini.

• Confirm or override the default model via Settings → NV oOS → Default Model (gpt-4.1 ships as the default).

• Set a default Gemini model under Settings → NV oOS → Default Gemini Model when Gemini is enabled.

• Choose the default provider from Settings → NV oOS → Default Provider so new assistants know whether to use OpenAI or Gemini by default.

• Adjust the request timeout under Settings → NV oOS → Request Timeout (minimum 5 s, default 30 s) to match your hosting environment.

• Select a default assistant with Settings → NV oOS → Default Assistant so REST and shortcode requests have a fallback.

• Decide on logging with Settings → NV oOS → Enable Logging when you need verbose diagnostics.

• Monitor token usage in Settings → NV oOS → Token Usage Statistics to track API consumption across users, providers, and models for billing and budget management.

• Choose your uninstall behaviour via Settings → NV oOS → Remove Data on Uninstall if this site should purge assistants and settings during cleanup.

• Configure Crawl4AI access in Settings → NV oOS → Tools when you want the Crawl4AI tool to be available to assistants.

• Review attachment MIME overrides in Settings → NV oOS → Attachments before enabling file uploads for end users.

• Review Send Group Email permissions in Settings → NV oOS → Tools to choose the capability and recipient cap for the group email automation.【F:includes/admin/class-wp-mcp-ai-admin-settings.php†L348-L359】【F:includes/admin/class-wp-mcp-ai-admin-settings.php†L938-L953】

• Connect Gmail under Settings → NV oOS → Tools → Connections → Gmail to enable Gmail search tools with OAuth 2.0. See Google OAuth Setup Guide for complete configuration steps.

• Connect QuickBooks Online under Settings → NV oOS → QuickBooks Company ID / API Key so the bundled reporting tool can fetch finance statements for authorised operators.【F:includes/admin/class-wp-mcp-ai-admin-settings.php†L906-L955】

• Configure Mailjet credentials in Settings → NV oOS → Mailjet API Key / Secret / From Email / From Name before enabling Mailjet-powered tools or Elementor widgets that send email on behalf of assistants.【F:includes/admin/class-wp-mcp-ai-admin-settings.php†L1008-L1054】

• Enable Federation & Discovery (Optional) in Settings → NV oOS → Federation & Discovery to publish your site's AI capabilities via /.well-known/ai-peer and optionally run a directory service for peer discovery. Configure regions, data tags, and rate limits to control how your site participates in the decentralized AI network.【F:docs/features/federation/federation-discovery.md†L1-L511】【F:FEDERATION-IMPLEMENTATION-SUMMARY.md†L1-L381】

• Configure Root Security Key (Optional) by adding define( 'WP_MCP_AI_ROOT_SECURITY_KEY', 'your-secure-key' ); to wp-config.php. This provides an additional security layer that can be enabled during emergency shutdown to require authentication before re-initializing the plugin.【F:docs/features/security/root-security-key.md†L1-L511】

• Enable Pro Dashboard (Optional) by adding define( 'WP_MCP_AI_PRO_DASHBOARD_ENABLED', true ); to wp-config.php. This activates the dedicated Pro Dashboard with ISO/IEC 27001 compliance monitoring, reporting, and management tools. See Pro Dashboard Documentation for details.

🧠 Language Model Providers (OpenAI, Gemini, Anthropic, Baseten, DeepSeek, OpenRouter, Kimi, DigitalOcean, NVIDIA NIM, Ollama, LM Studio, Hugging Face, Cloudflare)

A dedicated router transparently forwards chat completions to the active provider, allowing each request to target OpenAI, Gemini, Anthropic, DeepSeek, OpenRouter, Baseten, Kimi, DigitalOcean, NVIDIA NIM, a local Ollama instance, LM Studio, Hugging Face, or Cloudflare Worker AI while sharing the same assistant UX.【F:includes/class-wp-mcp-ai-language-model-router.php†L12-L86】 Configure the required API keys, default models, and the global default provider in Settings → NV oOS so new assistants inherit sensible defaults and administrators can switch providers without code changes.【F:includes/admin/class-wp-mcp-ai-admin-settings.php†L124-L333】【F:includes/admin/class-wp-mcp-ai-admin-settings.php†L505-L530】 Assistants can still override provider, model, and generation parameters on a per-post basis.

Privacy & Terms: All AI providers have specific terms and privacy policies:

• OpenAI: Terms | Privacy

• Google Gemini: Terms | Privacy

• Anthropic: Terms | Privacy

• NVIDIA NIM: Terms | Privacy

• Cloudflare: Terms | Privacy

• Hugging Face: Terms | Privacy

• Ollama/LM Studio: Self-hosted (no external data transmission)

• Baseten: Terms | Privacy

LM Studio Support

LM Studio with Function Calling - Full support for OpenAI-compatible function calling with local LM Studio instances:

• OpenAI-compatible message structure preserved for tool calls

• Tools/functions can be invoked by LM Studio models (e.g., qwen/qwen3-coder-30b)

• Streaming automatically disabled when tools are present for reliable execution

• Full backward compatibility with non-tool scenarios

• Connect via JSON-RPC endpoint (recommended) or SSE streaming

• See LM Studio setup guide for configuration details

Provider Priority List & Automatic Fallback

The plugin includes an intelligent provider priority system that automatically tries alternative providers when the primary one fails or is unavailable. In Settings → NV oOS, you can:

• Drag and drop providers to set your preferred order

• Automatic fallback - if the first provider fails, the system tries the next one

• Visual management - see all available providers (OpenAI, Gemini, Anthropic, Baseten, DeepSeek, OpenRouter, Kimi, DigitalOcean, NVIDIA NIM, Ollama, LM Studio, Hugging Face, Cloudflare) in one sortable list

• Flexible prioritization - adjust based on cost, performance, or availability needs

The first provider in the list serves as the default. If any provider returns an error, the router automatically attempts the next provider in the list until one succeeds. This ensures maximum uptime and resilience without manual intervention. All fallback attempts are logged for debugging and monitoring.

Local AI with Ollama

The Ollama provider enables privacy-focused, cost-free AI processing by connecting to a local Ollama or LM Studio instance running on your server or development machine. This is ideal for:

• Privacy-sensitive deployments where data must stay on-premises

• Development and testing without incurring API costs

• Custom or fine-tuned models not available through cloud providers

• Air-gapped environments without internet access

To configure Ollama:

1. Install Ollama on your server or local machine

2. Pull a model (e.g., ollama pull llama2)

3. Navigate to Settings → NV oOS → Ollama Configuration

4. Enter your Ollama endpoint URL (default: http://localhost:11434)

5. Click "Test Connection" to verify connectivity

6. Click "Fetch Models" to see available models

7. Select a model from the list or manually enter a model name

8. Set "Default Provider" to "Ollama (Local AI)" if you want it as the system default

The Ollama client supports the standard chat completion flow and automatically normalizes responses to match the OpenAI format for downstream compatibility. Note that some advanced features like tool calling may vary depending on the specific Ollama model you're using.

OpenAI model coverage

The plugin ships with presets for OpenAI’s current Responses, Reasoning, Audio, and Image APIs so site owners can choose the right model for each workflow. Token windows describe the maximum request size (messages, attachments, and tool payloads) the OpenAI API will accept for that model, while output limits reflect the largest single response the service will stream back. Leave a safety margin below each ceiling so assistants can add system instructions, tool calls, and knowledge snippets without hitting provider limits.