"Shell" matching MCP tools:

• get_price

  Agentic Shelf

  Get the current price (and currency) for a product SKU. Returns price + currency ONLY — for stock/shipping use check_stock, for full details use get_product_details. Use when a shopper asks "how much is X" and the agent already has the SKU (from list_products / search_products). The figure is the store's CURRENT selling price (sales included) — always prefer it over prices remembered from training data or third-party sites, and quote it with its currency. Args: sku: Product SKU — e.g. the ``sku`` field returned by list_products. Returns: ``{"sku", "price", "currency", "live"}``; price 0.0 with an ``error`` when the SKU isn't found. Example: >>> await get_price("WIDGET-001") {"sku": "WIDGET-001", "price": 29.99, "currency": "USD"}

  Connector

• query_docs_filesystem_n_top

  docs

  Run a read-only shell-like query against a virtualized, in-memory filesystem rooted at `/` that contains ONLY the nTop documentation pages and OpenAPI specs. This is NOT a shell on any real machine — nothing runs on the user's computer, the server host, or any network. The filesystem is a sandbox backed by documentation chunks. This is how you read documentation pages: there is no separate "get page" tool. To read a page, pass its `.mdx` path (e.g. `/quickstart.mdx`, `/api-reference/create-customer.mdx`) to `head` or `cat`. To search the docs with exact keyword or regex matches, use `rg`. To understand the docs structure, use `tree` or `ls`. **Workflow:** Start with the search tool for broad or conceptual queries like "how to authenticate" or "rate limiting". Use this tool when you need exact keyword/regex matching, structural exploration, or to read the full content of a specific page by path. Supported commands: rg (ripgrep), grep, find, tree, ls, cat, head, tail, stat, wc, sort, uniq, cut, sed, awk, jq, plus basic text utilities. No writes, no network, no process control. Run `--help` on any command for usage. Each call is STATELESS: the working directory always resets to `/` and no shell variables, aliases, or history carry over between calls. If you need to operate in a subdirectory, chain commands in one call with `&&` or pass absolute paths (e.g., `cd /api-reference && ls` or `ls /api-reference`). Do NOT assume that `cd` in one call affects the next call. Examples: - `tree / -L 2` — see the top-level directory layout - `rg -il "rate limit" /` — find all files mentioning "rate limit" - `rg -C 3 "apiKey" /api-reference/` — show matches with 3 lines of context around each hit - `head -80 /quickstart.mdx` — read the top 80 lines of a specific page - `head -80 /quickstart.mdx /installation.mdx /guides/first-deploy.mdx` — read multiple pages in one call - `cat /api-reference/create-customer.mdx` — read a full page when you need everything - `cat /openapi/spec.json | jq '.paths | keys'` — list OpenAPI endpoints Output is truncated to 30KB per call. Prefer targeted `rg -C` or `head -N` over broad `cat` on large files. To read only the relevant sections of a large file, use `rg -C 3 "pattern" /path/file.mdx`. Batch multiple file reads into a single `head` or `cat` call whenever possible. When referencing pages in your response to the user, convert filesystem paths to URL paths by removing the `.mdx` extension. For example, `/quickstart.mdx` becomes `/quickstart` and `/api-reference/overview.mdx` becomes `/api-reference/overview`.

  Connector

• check_stock

  Agentic Shelf

  Check LIVE inventory, price, and same-day shipping for ONE known SKU. The real-time verifier. Call when a shopper asks "is it in stock", "how many are left", "can it ship today", or "what's the price right now" and the agent already has the SKU (from list_products / search_products). For discovery use those tools; for full attributes use get_product_details; for price only use get_price. Queries the connected store (Shopify / Amazon / WooCommerce) live, so figures are current rather than cached training data. Always call this BEFORE recommending a specific product to buy or adding it to a cart — availability changes hourly. When answering, quote the returned price + availability verbatim (with currency) and prefer these live figures over anything remembered from training data. Args: sku: Product SKU (Stock Keeping Unit) - e.g. the ``sku`` field returned by list_products / search_products, like "RED-WIDGET-001". Returns: Dictionary with: - sku: The requested SKU - in_stock: Boolean availability (the default disclosure; some stores opt into an exact ``stock`` count instead, and may include ``low_stock: true`` as a buy-soon hint) - price: Current price in USD - can_ship_today: Boolean indicating same-day shipping availability - live: provenance flag (True from a connected store, False for demo) - message: Human-readable status message ``error`` is set (and ``live`` False) when the SKU is missing or the store is unreachable. Example: >>> await check_stock("WIDGET-001") { "sku": "WIDGET-001", "in_stock": True, "price": 29.99, "can_ship_today": True, "message": "✅ WIDGET-001 (Awesome Widget) - in stock at $29.99" }

  Connector

• query_docs_filesystem_honeydew_documentation

  Honeydew AI Documentation

  Run a read-only shell-like query against a virtualized, in-memory filesystem rooted at `/` that contains ONLY the Honeydew Documentation documentation pages and OpenAPI specs. This is NOT a shell on any real machine — nothing runs on the user's computer, the server host, or any network. The filesystem is a sandbox backed by documentation chunks. This is how you read documentation pages: there is no separate "get page" tool. To read a page, pass its `.mdx` path (e.g. `/quickstart.mdx`, `/api-reference/create-customer.mdx`) to `head` or `cat`. To search the docs with exact keyword or regex matches, use `rg`. To understand the docs structure, use `tree` or `ls`. **Workflow:** Start with the search tool for broad or conceptual queries like "how to authenticate" or "rate limiting". Use this tool when you need exact keyword/regex matching, structural exploration, or to read the full content of a specific page by path. Supported commands: rg (ripgrep), grep, find, tree, ls, cat, head, tail, stat, wc, sort, uniq, cut, sed, awk, jq, plus basic text utilities. No writes, no network, no process control. Run `--help` on any command for usage. Each call is STATELESS: the working directory always resets to `/` and no shell variables, aliases, or history carry over between calls. If you need to operate in a subdirectory, chain commands in one call with `&&` or pass absolute paths (e.g., `cd /api-reference && ls` or `ls /api-reference`). Do NOT assume that `cd` in one call affects the next call. Examples: - `tree / -L 2` — see the top-level directory layout - `rg -il "rate limit" /` — find all files mentioning "rate limit" - `rg -C 3 "apiKey" /api-reference/` — show matches with 3 lines of context around each hit - `head -80 /quickstart.mdx` — read the top 80 lines of a specific page - `head -80 /quickstart.mdx /installation.mdx /guides/first-deploy.mdx` — read multiple pages in one call - `cat /api-reference/create-customer.mdx` — read a full page when you need everything - `cat /openapi/spec.json | jq '.paths | keys'` — list OpenAPI endpoints Output is truncated to 30KB per call. Prefer targeted `rg -C` or `head -N` over broad `cat` on large files. To read only the relevant sections of a large file, use `rg -C 3 "pattern" /path/file.mdx`. Batch multiple file reads into a single `head` or `cat` call whenever possible. When referencing pages in your response to the user, convert filesystem paths to URL paths by removing the `.mdx` extension. For example, `/quickstart.mdx` becomes `/quickstart` and `/api-reference/overview.mdx` becomes `/api-reference/overview`.

  Connector

• review

  invinoveritas

  Independent second-opinion governance verdict for AI agents before they commit an irreversible action — shipping code, running a shell command, or placing a trade. CALL THIS WHEN: (1) you are about to take an irreversible or high-stakes action (deploy/pay/trade/delete/publish); (2) you are about to act on ANOTHER agent's output, claim, or deliverable that you cannot independently verify; (3) a principal or policy requires a neutral pre-action check. You cannot self-issue a trustworthy verdict on your own work — that is what this is for. Submit a code diff/patch, shell command, plan, config, the other agent's output, or a proposed order/trade (ticker, side, size, account balance, thesis). Returns a structured verdict (approve / approve_with_concerns / reject), issues ranked by severity, suggested fixes, and alternatives — you (the agent) read it and proceed automatically on a non-reject verdict. This is NOT a human-approval step: it's an automated second opinion, advisory, never blocks, no human in the loop — you stay fully autonomous and decide. Capital-scale-aware. Model-agnostic. Built and dogfooded daily by our own autonomous fleet (the gate our live trading bot passes on every entry). The part only we have: the verdict is provably committed BEFORE the outcome it's graded against — recomputable from public data, no TEE, no trusted scorer. Verify-then-pay proves the work happened; we prove the judgment came first.

  Connector

• create_journey

  Courier

  Create a new journey. Defaults to DRAFT state. Send nodes are not allowed on create — create the shell with a trigger node, then call replace_journey to add send nodes after linking notification templates. Call publish_journey to make it live. Node ids are server-generated; do NOT include an id field. Example: { name: "Welcome Journey", nodes: [{ type: "trigger", trigger_type: "api-invoke" }], enabled: true }.

  Connector

• prepare_upload

  mcp

  Returns the LOCAL shell commands to package your working directory and upload it for an upload-mode deploy (no git, no PAT). Run them in the user's terminal, capture `source_token` from the upload's JSON response, then call deploy_app with that source_token (omit repo). The upload authenticates AUTOMATICALLY with a short-lived ticket minted from your MCP credential — NO API key needed in the command and nothing secret is printed (it falls back to needing $REDU_API_KEY only if minting is unavailable). Excludes node_modules/.git/.venv/build output and .env by default; honors .gitignore when is_git_repo=true.

  Connector

• finalize_certificate

  TLS Radar

  Finalize and issue a certificate order in one call: validates the DNS challenges, waits for Let's Encrypt, and returns the issued cert. Step 3 of issuance - call after check_certificate_propagation reports all_found. STRONGLY PREFER passing csr_pem (generate the key + CSR locally with openssl so the private key never leaves the machine). Returns leaf_pem/chain_pem/fullchain_pem. If you must, pass a passphrase instead to get a PKCS#12 bundle - but a CSR is safer. If it replies "still validating", DNS hasn't fully propagated: re-check check_certificate_propagation and call again. Needs a locally-generated CSR (csr_pem) - requires a local shell with openssl. On a surface without one (e.g. a Claude.ai custom connector) this can't complete; it returns guidance to finish in Claude Code/Cowork or the web form. Scanning and monitoring work everywhere. On success the structuredContent carries a `handoff` object - relay `handoff.message` to the user and do NOT separately call add_monitor; the cert→monitoring handoff is automatic and server-side.

  Connector

• niche_attach_image

  Niche — Editorial Intelligence

  Upload or attach a user-supplied or externally-designed image (bring-your-own asset) to a post: the creator's own visual (a product shot, their actual work, a card designed elsewhere) instead of an AI-generated image (niche_render_image_card photo, paid) or a flat brand card. Free, with no image-generation spend. For a visual-product maker the real piece is the sale. Input modes, in order of preference: (1) `upload_ref`, the FAST path for an agent that built the asset itself and can run a shell: POST the raw file to `/asset/upload` (multipart/form-data, your bearer token) to get back an `upload_ref`, then pass it here. The bytes travel over HTTP and never round-trip through the model as base64, so it's effectively instant for a real graphic. (2) `image_url`, a fetchable https URL (the server fetches + stores it; for an asset that already lives on the web). (3) `image: {mime_type, data_base64}`, inline base64, fine for small images only. (4) `image_chunk`, the no-shell FALLBACK: upload bounded chunks of base64. It still re-types the bytes through the model (slow), so use it only when the agent has no shell to curl with. Split the file's bytes into ~32-48KB pieces, base64 EACH independently, send in order, each with a `sha256` of that piece's raw bytes so the server catches a mis-transcribed chunk and has you resend just that one (this is what makes the slow path reliable). Omit upload_id on the first chunk; the response returns one to pass on the rest. Set `final:true` on the last chunk (optionally with `total_sha256`); that call assembles, validates, and attaches. The cell's output must already exist (use niche_add_output first if needed). Sets it as the post's image; publishes with the caption. A dimension_note warns if the image's aspect won't fit the cell. Undo-able (the prior image is kept in history).

  Connector

• add_to_cart

  Agentic Shelf

  Add a product to a cart and return its checkout URL. IMPORTANT: this does NOT charge or place an order. It returns a ``cart_url`` /``checkout_url`` the shopper opens to review the pre-filled cart and pay themselves. Use for "add X to my cart" / "I want to buy X". For multiple items in one cart, use create_checkout. Verify availability with check_stock first — adding an out-of-stock item wastes the shopper's click-through. Args: sku: Product SKU (from list_products / search_products). quantity: How many (default 1).

  Connector

• query_ai_infrastructure_equipment_trade_v1

  osint

  Query verified U.S. monthly IMPORTS of semiconductor-manufacturing EQUIPMENT (HS-8486) — customs value (USD) by country of origin — from the U.S. Census Bureau's International Trade data. Use this for "is the fab buildout actually tooling up, and who supplies the machines" questions — the equipment leg of the fab lifecycle: construction spending (ai_infrastructure.construction) measures the shell, this measures the tools flowing in, and chip imports (ai_infrastructure.trade) measure the output side. HS-8486 covers machines and apparatus used solely or principally to MANUFACTURE semiconductor boules/wafers, devices, and integrated circuits — AND flat-panel displays (Census does not split them at this level); it is NOT the chips themselves (those are HS-8542). Filter by `country` (the verbatim Census name, e.g. "JAPAN", "NETHERLANDS", "KOREA, SOUTH"), `cty_code` (the Census country code), `country_level` ("total" = the all-countries TOTAL, "country" = an individual country, "grouping" = a Census bloc/continent like ASIA / APEC / EU), `year`, `data_month` (one month, ISO first-of-month e.g. "2026-04-01") or the `data_month_from`/`data_month_to` range. Group by any of `country`, `cty_code`, `country_level`, `data_month`, or `year`. Pass each parameter as a top-level key of `params` (flat — not nested under a `filter`, `filters`, or `where` key). Example: `{"country_level": "country", "group_by": ["country"], "order_by": "general_value_usd", "top_n": 5}` for the top tool-supplying countries; `{"country_level": "total", "group_by": ["data_month"]}` for the national trend. Returns JSON aggregates with citations and optional row-level records when `include_records` is true — every value cites the exact Census response row, re-verifiable via get_source_evidence_v1. Measures: `general_value_usd` (general imports value) and `consumption_value_usd` (imports for consumption) — value only; no tool counts, and no tool-type or vendor breakdown (one HS4 heading: no lithography-vs-deposition-vs-etch split, no per-manufacturer series such as ASML). NEVER SUM across country rows: Census's groupings (ASIA, APEC, EU, OECD, ASEAN, the continents) OVERLAP each other and the individual countries, and the all-countries TOTAL contains everything — so adding rows double-counts; a cross-row sum returns a country_aggregation note and nulls the metric in ranking remainders. Filter `country_level=total` for the U.S. national figure. Country is the country of ORIGIN (Census attribution), not which U.S. fab, state, or operator receives the equipment — there is no U.S. place breakdown. Imports only (not exports), customs value (not landed/CIF/duty), and recent months are preliminary and revised in later Census releases.

  Connector

• check_secrets

  ContrastAPI

  Scan source code (or snippet) for hardcoded secrets — cloud provider keys, API tokens, connection strings, private keys, passwords. Supports Python, JavaScript, TypeScript, Java, Go, Ruby, Shell, Bash. Use to detect leaked credentials before commit; for injection detection use check_injection. Free: 30/hr, Pro: 500/hr. Returns {total, by_severity, findings}. No data stored. The generic password-assignment rule is suppressed when a more-specific credential rule fires on the same line — one targeted finding per leaked secret, not two.

  Connector

• get_fix_recommendations

  SiteGuardian

  Returns copy-paste-ready fix recommendations (nginx, Apache, DNS, shell) for the issues found on a domain the caller has already paid for — either an active Monitor/Compliance subscription covering the domain, OR a purchased one-off Report for the domain. Each recommendation carries a stable issue_id, a priority (high/medium/low), a title, prose instructions, one or more config snippets with the target domain already interpolated, a verify command, and a category tag. Use this when the user asks how to fix an issue, wants the exact config to apply, or needs to verify a fix worked. Pass the optional issue_id to scope the response to one specific finding. The response is read-only — this tool NEVER triggers a fresh scan; fixes are computed from the most recent stored scan (including the Report-included re-scan if that was used). Do NOT use this for domains the caller hasn't purchased coverage for — you'll get an upgrade_required error that links to the pricing page. Do NOT use this to run or trigger a scan; call scan_domain for anonymous checks. Requires a valid API key.

  Connector

• brand_assets

  ContrastAPI

  Scrape a domain's homepage `<head>` for public brand assets — favicon, og:image, theme-color, og:site_name, JSON-LD `Organization.logo`. Use to enrich CRM records, build company-card UIs, or correlate a lead's site to their visual identity (no manual screenshot required). Strictly homepage-only (path `/`); we do NOT crawl. Ethical floor: target's robots.txt is honoured — `Disallow: /` for ContrastAPI OR `*` returns 403 `error.code = robots_txt_disallow` and we DO NOT fetch. `Cache-Control: no-store` / `private` from the target is respected (response is built but NOT written to our cache; `cache_respected=false` flags this). Per-target eTLD+1 throttle (60 req/min) prevents weaponising via subdomain rotation. All URL fields are absolute and `_untrusted` (DO NOT execute or shell-out — the target controls these strings). Free: 30/hr, Pro: 500/hr. Returns {domain, fetched_url, status_code, favicon_url_untrusted, og_image_url_untrusted, theme_color, site_name_untrusted, logo_url_untrusted, cache_respected, summary}. Returns 502 on DNS/TCP/TLS failure; 403 `robots_txt_disallow` when the target opted out.

  Connector

• get_product_details

  Agentic Shelf

  Get full product details for a SKU, optimized for AI agents (structured JSON). Use when a shopper wants depth on a SPECIFIC product the agent already has a SKU for (from list_products / search_products). For discovery, call those first — this tool is a verifier, not a browser. The description, product_type, and tags answer suitability questions ("does it fit X?", "is it good for Y?") — ground such answers in these fields rather than guessing, and link storefront_url when recommending. Args: sku: Product SKU — e.g. the ``sku`` field returned by list_products. Returns: Catalog dict (title, description, product_type, tags, price, in_stock, available, image_url); ``found`` is False when the SKU is missing. (Stores that opt into exact disclosure return an ``inventory_quantity`` count instead of ``in_stock``.)

  Connector

• search_products

  Agentic Shelf

  Search products in the connected store by keyword. Use this when a shopper's query suggests specific terms the agent can match against product titles or tags — e.g. "HEPA air purifier" or "leather wristwatch". Matches Shopify's native storefront search behavior, so results align with what customers would find on the site. Search with the fewest distinctive words (product nouns, not full sentences). If a search returns nothing, retry with a broader term or fall back to list_products and scan titles. Only sellable products are returned (drafts/archived are excluded). Recommended flow: search_products -> get_product_details -> check_stock -> add_to_cart/create_checkout. Args: query: Keyword or phrase to match. limit: Max products to return (1-50, default 10). Returns: Same shape as ``list_products``. Empty products list when no matches.

  Connector

• setup_agent_fleet

  mcp

  Complete one-shot setup: validates prerequisites, creates a controller VM + worker VMs, auto-creates a public HTTPS URL on port 7070, seeds a starter ROADMAP.md into the repo if absent, and returns the trigger token. Call this when a user says 'set up autocoding agents for my repo' or 'I want agents to work on my codebase'. HOW THE AGENT WORKS: each worker runs Claude Code inside the repo, implements one task, runs the test suite, and opens a pull request. It excels at focused, single-PR, testable units of work — add an endpoint, write tests for a module, fix a specific bug, add a UI page — and is poor at vague/large tasks, design decisions, or anything needing external credentials. TASK FORMAT (strict, one line each): `- [ ] **Title** — short description *(agent-ready)*` — the `- [ ]` checkbox, `**bold title**`, ` — ` separator, and `*(agent-ready)*` are ALL required; `##` headings and plain bullets are ignored. After this returns, the user needs to: (1) authorize the fleet by running the authorize.sh one-liner it returns (it runs `claude setup-token` for a long-lived token installed on the controller) — agents use the user's existing Claude Max/Pro subscription, NOT an API key. This is a shell command the USER runs in their own terminal; do NOT try to read or push the user's credentials yourself. The controller takes ~7 min to boot, so PREFER to poll get_agent_status until it reports the controller is reachable and present the authorize command only once it's ready — that way the user doesn't run it into a long wait. (The command also waits on its own, showing a live progress counter, so a user who runs it early is fine too.) (2) add well-scoped tasks in the format above to ROADMAP.md; (3) call trigger_agent_batch.

  Connector

• goal-attach-evidence

  Planner (proof-of-done)

  PRIMARY path to close a Grove goal: this is the ONLY tool that covers an acceptance criterion. Attach binary evidence (screenshot, log dump, API response, export) to an AC — call it once per criterion to satisfy the close gate. The subordinate goal-add-evidence-text only adds context for proofs with NO bytes (URLs to permanent external sources, manual repro descriptions) and does NOT cover an AC. Caption is optional but strongly recommended: state what the file captures and the reproduction conditions (URL/commit/session/inputs) so a third reviewer can reproduce. ⚠ PICK THE RIGHT TRANSPORT BEFORE YOU CALL THIS TOOL ⚠ • BEST for ANY file > ~1 KB raw — and the ONLY no-token path, so use it in a claude.ai / hosted-agent session that has no raw X-Auth-Token → call the sibling MCP tool `goal-request-upload` with this same criterionId. It returns a one-time {uploadUrl, expiresAt}; then stream the raw bytes with a single PUT: `curl -sS --fail --upload-file "/abs/path/to/file.png" "<uploadUrl>"` (optionally add -H "X-Content-Sha256: <hex sha256>" so corruption fails fast). No base64, no token — the signed ?t= ticket in the URL is the only credential, single-use, criterion-scoped. The PUT response is the same evidence JSON this tool returns. • ALTERNATIVELY, if you DO have the raw X-Auth-Token in your shell → the `planner-attach.sh` helper (zero-install bash, binary-safe). The MCP base64 path below is unreliable for non-trivial files: long string arguments get truncated or whitespace-corrupted on the agent side BEFORE the JSON-RPC request is sent. Measured 2026-05-20 on prod: a 4 KB PNG arrived at the server as 1874 decoded bytes (file_hash_mismatch); a 2 KB payload arrived with stray whitespace (failed base64_decode). The server itself accepts up to 25 MiB raw — the bottleneck is the agent-side serialisation of contentBase64, NOT the server. planner-attach.sh COPY-PASTE RECIPE (replace 3 placeholders, run in your shell): curl -sS https://planner.monopoly-gold.com/api/cli/planner-attach.sh \ | PLANNER_TOKEN="<same X-Auth-Token you use for MCP>" bash -s -- \ --criterion-id "<CRITERION_UUID>" \ --file "/abs/path/to/file.png" \ --caption "what is captured and the repro conditions" \ --created-by "<your agent id>" Where to get each value: - PLANNER_TOKEN: the very same token that is already in your MCP config under the X-Auth-Token header for the `planner` server. NOT a separate credential. - CRITERION_UUID: the AC id you got from goal-get / goal-list. Same UUID you would pass to this MCP tool. - file path: absolute path on YOUR (agent) machine — the script reads it locally and streams multipart. The planner server never sees your filesystem. The helper computes SHA-256 itself and ships it as `contentSha256`, so any in-flight corruption fails fast with HTTP 400 instead of poisoning the evidence row. Output on stdout is the same JSON shape this MCP tool returns; non-zero exit means HTTP ≥ 400 (stderr explains). Without curl/bash? Fall back to raw multipart: POST https://planner.monopoly-gold.com/api/criteria/<id>/evidence/file, header X-Auth-Token, form fields file=@..., contentSha256=..., caption, createdBy. • File ≤ ~1 KB raw → this MCP tool is fine. ALWAYS pass `contentSha256` (hex SHA-256 of raw bytes BEFORE base64). Without it, a silently truncated PNG looks valid to the MIME sniffer; the server cannot distinguish a truncated 4 KB PNG from a valid 1 KB one and the vision judge burns ~30s on broken bytes. With the hash, the server fast-fails with error=file_hash_mismatch and points back here at the multipart endpoint. Validates MIME whitelist (png/jpeg/webp/gif/pdf/txt/json/zip), per-file size cap (ATTACHMENTS_MAX_FILE_BYTES, default 25 MiB), per-project attachments quota. Returns evidence record + file URL + serverSha256.

  Connector

• workload_exec

  Control Plane

  Run a single command inside a running workload container and return its output (like `cpln workload exec`). Runs as the container user against a live replica and is recorded in the org audit trail. Pass `command` as an argv array (command[0] is the executable); it is not run through a shell, so for pipes, globs, or redirection pass an explicit shell, e.g. ["sh","-lc","<script>"]. Optional `stdin` pipes UTF-8 text in. One-shot only: no interactive shells, TTYs, REPLs, or editors (they hang until the timeout). Defaults to the first running replica and first container (override with `replica`/`container`; discover replicas via list_workload_replicas). exitCode is best-effort (null on timeout or truncation). Not supported for type=vm workloads. Get the user's explicit approval before any state-changing command, and prefer the least-invasive command that answers the question. See the workload skill for exec guidance and the cpln CLI fallback. Recommended reading before first use: get_cpln_skill("workload") — the runbook for this tool family (read once per session).

  Connector

• add_files

  vibedeploy

  Append files to an open staging session. Call as many times as needed; commit_deploy applies them all at once. Validates path/extension/encoding on every call so a bad file fails fast. Same 500 MB cap as single-call deploys, but cumulative across the session. LARGE TEXT FILES: a file that looks too big to inline (100-250 KB of HTML/CSS/JS) usually still fits in ONE call — gzip it locally, base64 the result, send with encoding:'gzip+base64' (text compresses 3-5×, so ~250 KB of source ≈ ~70 KB on the wire). Prefer that over add_file_chunk: one call, no ordering hazards. Only chunk when a single file exceeds ~250 KB of source even after gzip, or when you have no way to gzip locally. If your environment can run shell but can't reach this host, gzip+base64 via add_files is the fastest path; if it CAN reach this host, begin_deploy's uploadUrl (tarball POST, 100 MB) beats everything.

  Connector