Skip to main content
Glama
GodPrompt MCP Server

GodPrompt MCP Server

GodPrompt MCP Server

A Model Context Protocol (MCP) server for GodPrompt — AI software-development workflow guidance with task routing, TDD, debugging protocols, verification gates, and progressive disclosure.

MCPVault: verified

Agent Harness Review — US$49

GodPrompt offers a bounded Agent Harness Review for developers and small teams whose coding agents lose context, overstep scope, skip verification, or repeat mistakes.

GodPrompt Agent Harness Review — US$49 one-time. For one repository, send up to three agent-control artifacts (AGENTS.md, CLAUDE.md, project rules, or equivalent) plus one short failure example. The review returns five priority risks, one proposed revised instruction block or patch, and a next-task verification checklist, with a target of 24 hours after payment and usable inputs.

To buy the review, use the US$49 one-time GitHub Sponsors tier. GitHub sign-in is required. After sponsoring, email info@tomiseregi.si with subject [GP-RS1] Agent Harness Review and include the GitHub username used for the sponsorship plus the review inputs. Delivery begins after the US$49 sponsorship is independently verified for this review. The free MIT GodPrompt MCP remains unchanged and free. See a transparent sample deliverable before deciding; it is illustrative, not a client result or benchmark. Full frozen experiment terms.

Before emailing, sanitize your review inputs: remove passwords, API keys, access tokens, recovery codes, regulated or unrelated personal data, and unrelated private repository content. Do not send .env files, full private repositories, or unredacted logs that might expose credentials. Share only the minimum sanitized instruction excerpts and failure example needed for the review; if those are insufficient, the scope may need to be narrowed or the review declined under the linked terms.

Related MCP server: prompts-mcp-server

Tools

Tool

Description

get_god_prompt

Full GodPrompt.md single-file payload (~40KB)

get_core_skill

SKILL.md — core protocol (load when relevant, ~10KB)

get_protocols

references/01-PROTOCOLS.md — deep execution guides (~13KB)

get_gates

references/02-GATES.md — verification checklists (~9KB)

get_anti_patterns

references/03-ANTI-PATTERNS.md — red flags & recovery (~9KB)

classify_task

Deterministically route a task to one of 9 GodPrompt task types, or UNCLASSIFIED when no reliable signal exists

get_version

Version info and server metadata

create_work_contract

Create a portable, fingerprinted task contract with explicit scope, acceptance criteria and required verification commands

review_work_evidence

Fail-closed review of reported changed paths, verification receipts, acceptance results and forbidden actions

review_code_change

Inspect a supplied unified diff for file/line-anchored review signals; no repository access or semantic correctness claim (v1.2.0+)

audit_requirements

Audit supplied requirements for missing acceptance scenarios, duplication and ambiguous criteria (v1.2.0+)

build_test_matrix

Draft positive, negative, boundary and regression cases linked to supplied requirements; does not run tests (v1.2.0+)

get_tailored_guidance

Retrieve bounded protocol/quality guidance for a task type and phase instead of loading complete reference files (v1.2.0+)

diagnose_failure

Classify caller-supplied symptoms into testable hypotheses without echoing raw logs or asserting root cause (v1.2.0+)

assess_release_readiness

Find blockers in reported CI, artifact provenance, rollback, migrations and approvals; never authorizes deployment (v1.2.0+)

create_context_handoff

Produce bounded cross-chat/agent state with best-effort secret redaction and explicit verification limitations (v1.2.0+)

compare_evaluation_runs

Compare caller-supplied paired cases and observed pass rates, durations and costs with sample-size warnings (v1.2.0+)

audit_agent_trajectory

Review bounded, sanitized agent event metadata for repeated failures, scope drift, stale checks and unsupported completion claims (v1.3.0+)

trace_change_impact

Use supplied dependencies, requirement mappings and test receipts to trace affected files and uncovered verification (v1.3.0+)

review_mcp_tool_contracts

Flag possible MCP schema, annotation and behavioral contract inconsistencies without invoking tools (v1.3.0+)

audit_agent_instructions

Inspect scoped caller-supplied instruction excerpts for duplicate rules, opposing constraints and suspicious bypasses (v1.3.0+)

design_adversarial_tests

Suggest bounded negative, boundary, property, metamorphic and trust-boundary cases; never run tests (v1.3.0+)

Version 1.1.1: create_work_contract and review_work_evidence expand the seven original MCP tools to nine. Older hosted or version-pinned installations may still expose only the original seven until separately updated. Both tools remain stateless and read-only, and neither independently authenticates caller-supplied evidence. See the task-contract workflow and security limitations.

Version 1.2.1: All seventeen public MCP tools are available, including the eight engineering-quality functions introduced in v1.2.0. This patch refines code-review signal precision, requirement-conflict detection and duplicate CI-check blocking without adding permissions or storage. The Codexify checkpoint remains an optional stdio-only eighteenth tool. Install a versioned MCP package or inspect the hosted tools/list response to determine which functions your client actually receives. None of the tools independently execute tests, inspect external repositories or authorize deployment. See the toolkit QA and limitations. The OpenAI marketplace distribution remains a separate, Skill-only integration until its eventual dedicated upgrade.

Version 1.3.0: All 22 public read-only tools are available from the published hosted MCP and npm stdio package. Five P4 diagnostics were added, with strict bounded output schemas and paired evaluation statistics. The optional local Codexify checkpoint is the 23rd tool only when explicitly enabled on local stdio. The remote MCP does not execute code, fetch repositories or approve changes. See P4 evidence and release status. The combined OpenAI Skill+MCP source package is ready for review, but the existing directory listing is still Skill-only; GHCR 1.3.0 publication is a separately manual step.

Version 1.4.2 (P5 test-selection patch): The public tool count remains 22, with improved advisory trajectory diagnostics, optional accessible MCP Apps evidence display for two existing tools, and smaller task-specific guidance suggestions. The local npm distribution adds godprompt-repo-map to derive a bounded static JavaScript/TypeScript dependency map from an explicitly authorized checkout, and godprompt-proof pack to combine pinned GitHub evidence with optional Codexify receipts, invalidating contradictions. Both CLIs are local opt-in, not hosted MCP permissions. The free coding-agent benchmark runner and 12 deliberately failing grader microtasks help set up paired A/B evaluation; they do not prove a model productivity improvement. The v1.4.2 patch also resolves NodeNext source imports and detects more instruction-bypass wording; clients must update the local package for these improvements.

Optional local proof bridge (from v1.1.1)

The new godprompt-proof CLI (Node 22+, no additional dependencies) can normalize pinned GitHub CI checks and exported Codexify command receipts into inputs for the existing GodPrompt evidence reviewer. It also supports an optional single 256 KiB local JSONL feedback file for project-scoped generic lessons, on-demand read-only metrics, and versioned workflow previews from that same file (no telemetry or automatic rule changes). No database server, GPU, background worker or global telemetry is involved.

GitHub can be read directly or from a sanitized snapshot produced with an existing GitHub connector, with no new token required for that offline path. Codexify is client-orchestrated: the authorized coding agent produces a receipt export, not a direct remote connection from the public Worker. The hosted MCP has no access to any user's local feedback file. No automatic uploads, training or rule promotion occur.

The optional CLI ships with v1.1.1, but it is not installed or enabled automatically on a user's computer. The hosted MCP cannot read local files. See local proof bridge setup, limitations, and examples. The CLI is optional; the MCP server continues to work without it.

ChatGPT publishing and storage: The public hosted MCP remains stateless and does not create personal memory or store a JSONL file for ChatGPT users. Per-project learning is an explicitly enabled local CLI, not a hosted MCP feature. See the hosting, privacy, local installation and release checklist. Cloud sync is not included.

Progressive Disclosure

For minimum context usage in v1.2.0 and later, route the task with classify_task, then call get_tailored_guidance for just its current phase. Load get_core_skill when the full core workflow is necessary, and get_protocols, get_gates, or get_anti_patterns only for deeper guidance. Existing released MCP installations can continue starting with get_core_skill until their tool catalogs are updated. Use get_god_prompt only when the complete single-file text is actually needed.

Example queries

  • "Classify this refactor and tell me which GodPrompt workflow applies."

  • "Show me the verification gates I should satisfy before I claim this bug fix is complete."

  • "Load the debugging protocol for a failing integration test without loading the full GodPrompt."

Evaluation

GodPrompt's benchmark methodology, deterministic task corpus, evaluator logic, and reproducible run/export tooling live in the source project: GodPrompt Bench. No frozen reference-model run or raw benchmark artifacts have been published yet.

The MCP server does not run the benchmark or claim model-level superiority itself; it distributes the GodPrompt content evaluated by that suite.

Connect

Hosted remote (no Node.js required)

Use the public Streamable HTTP endpoint directly in MCP clients that support remote servers:

https://god-prompt-mcp.tomi-seregi99.workers.dev/mcp

The hosted endpoint exposes the read-only GodPrompt tools included in its currently deployed release and requires no API key or local Node.js runtime. An unreleased source branch can have a different tool list; always inspect the live tool catalog after deployment. It accepts both the deployed initialize flow and the current MCP 2026-07-28 server/discover flow so clients can connect while the ecosystem transitions between protocol eras. It is also published as a remote option in the Official MCP Registry alongside the installable stdio packages.

Agent Skills over MCP (SEP-2640)

On MCP 2026-07-28, both the hosted and stdio servers advertise the Final io.modelcontextprotocol/skills extension from SEP-2640. Supporting clients can discover the same relevance-scoped god-prompt Agent Skill with skills/list / skills/get, then fetch SKILL.md and its three reference files through resources/read. The skill manifest includes SHA-256 digests and exact byte sizes so clients can verify what they import. Clients that do not implement this extension continue to use the 22 public GodPrompt tools normally.

This transports GodPrompt's existing skill package; it does not force the skill into every request or change the host's own approval, verification, or activation policy. See the MCP Skills extension specification.

ChatGPT Plugin Directory (skill-only)

For the lowest-friction ChatGPT path, install GodPrompt from the first-party Plugin Directory. That published plugin carries the scoped god-prompt Agent Skill; use the custom MCP app setup below when you specifically want the 22 callable public GodPrompt tools.

ChatGPT (hosted remote MCP)

On ChatGPT web, supported Business and Enterprise/Edu workspaces can connect GodPrompt's hosted endpoint as a custom MCP plugin. Workspace permissions apply: Business admins/owners can create connections; Enterprise/Edu admins/owners and specifically authorized developers can create connections, while only admins/owners publish them for a workspace.

  1. Open Plugins, select +, then Add custom MCP server. The current Plugins flow does not require Developer Mode.

  2. Enter a name and https://god-prompt-mcp.tomi-seregi99.workers.dev/mcp as the public HTTPS MCP endpoint. Choose no authentication; GodPrompt requires no API key.

  3. Review the risk warning and select I understand and want to continue, then Create as a plugin.

  4. Review the discovered read-only GodPrompt tools and metadata (the count depends on the deployed release), then install the plugin if your workspace permits it. Approved ChatGPT tool snapshots may require an admin refresh after a release.

  5. Start a new ChatGPT chat, type @, select the installed GodPrompt plugin, and try a read-only tool. Workspace approval may be required.

Legacy custom-app setup: If your workspace still uses the older interface, open Settings > Apps > Create (or the workspace's app settings, if you are an admin/owner), use the same endpoint with no authentication, choose Scan Tools, then Create. This remains an alternative, not a prerequisite for the current Plugins flow.

ChatGPT connects to remote MCP servers rather than directly to local stdio servers. See OpenAI's Connecting custom MCP servers in ChatGPT for current permissions and setup steps.

npx -y god-prompt-mcp

The public npm package runs the local stdio server directly. It requires Node.js 22+ and no API key or account.

Windows custom stdio clients: If a client reports spawn npx ENOENT or spawn npx.cmd EINVAL, it may be trying to execute npm's Windows batch shim without a compatible launcher. Prefer the hosted endpoint, or, for an affected client that accepts a custom stdio command, use:

{
  "command": "C:\\Windows\\System32\\cmd.exe",
  "args": ["/d", "/s", "/c", "npx -y god-prompt-mcp"]
}

This is a client-launcher workaround, not a GodPrompt MCP server change. Keep existing working cross-platform npx configurations unchanged.

Docker / GHCR (local stdio; no host Node.js)

For MCP clients that can launch a local process, the public GHCR image runs the read-only tools of its pinned published version without requiring Node.js on the host. Docker with Linux-container support is required; no API key is needed.

{
  "command": "docker",
  "args": ["run", "--rm", "-i", "ghcr.io/akzar1el/god-prompt-mcp:1.1.1"]
}

Keep -i so the MCP client can exchange JSON-RPC over standard input/output; do not add -t (a TTY changes the stdio stream). The version tag selects the verified v1.1.1 image; update it deliberately for future releases rather than silently following the moving latest tag. The image is Linux/amd64 and runs as non-root user 10001:10001. The newer v1.3.0 Docker source has passed image and stdio smoke testing, but the GHCR version tag requires a separate explicit manual publish and post-publish verification. See the final GHCR checklist; do not change the pinned example before that verification. If Docker is unavailable, use the hosted remote endpoint instead.

Glama

GodPrompt MCP is published on Glama. Use the server page to inspect the tools and connect it to a supported MCP client.

Kiro

Install this repository as a native Kiro Power to get the portable GodPrompt Agent Skill and MCP server together. In Kiro, open Powers → Add Custom Power → Import power from GitHub, enter:

https://github.com/AKzar1el/god-prompt-mcp

Kiro reads the repository's Agent Plugins 1.0 plugin.json, skills/, and mcp.json; the bundled MCP server activates with the Power instead of requiring a separate user-level MCP entry.

If you only want the MCP server, use the direct install button:

Add to Kiro

Both paths ultimately launch the published npm package through npx. Node.js 22+ is required for the MCP server.

Visual Studio Code

VS Code supports Agent Plugins 1.0 packages directly, so the lowest-friction path installs GodPrompt's portable Agent Skill and bundled MCP server together:

  1. Run Chat: Install Plugin From Source from the Command Palette.

  2. Enter https://github.com/AKzar1el/god-prompt-mcp.

  3. Review the repository source before confirming the install. VS Code discovers the root plugin.json, skills/god-prompt, and mcp.json from the same package.

If you only want the Agent Skill, install it at user scope so VS Code discovers it from ~/.copilot/skills:

gh skill install AKzar1el/god-prompt-mcp skills/god-prompt --agent github-copilot --scope user

If you only want the callable MCP tools, use the native MCP install button:

Install in VS Code

The plugin and MCP-only paths both launch the published local stdio server through npx. Node.js 22+ is required; no API key is needed.

Amazon Q Developer

Amazon Q Developer supports both local process MCP servers and remote HTTP MCP servers. For the no-Node path, add GodPrompt to the mcpServers object in the Q CLI or Q Developer IDE agent configuration:

{
  "mcpServers": {
    "god-prompt": {
      "type": "http",
      "url": "https://god-prompt-mcp.tomi-seregi99.workers.dev/mcp"
    }
  }
}

The hosted endpoint is open and requires no API key. If you prefer local stdio, use the same Amazon Q MCP configuration shape with "command": "npx" and "args": ["-y", "god-prompt-mcp"]; that path requires Node.js 22+. See Amazon Q Developer's MCP guide and agent configuration format for the current CLI/IDE configuration locations and scope rules.

OpenAI Codex

GodPrompt can be used in Codex as either a portable Agent Skill or a native local MCP server.

Install the standards-compatible Agent Skill directly from this repository with Codex's built-in $skill-installer:

$skill-installer install https://github.com/AKzar1el/god-prompt-mcp/tree/main/skills/god-prompt

Or register the published stdio server with Codex's native MCP CLI:

codex mcp add god-prompt -- npx -y god-prompt-mcp

Run codex mcp list to verify the registration. The Agent Skill gives Codex the reusable workflow directly; MCP adds GodPrompt's progressive-disclosure tools. Node.js 22+ is required for the MCP server.

Gemini CLI

Install the repository as a native Gemini CLI extension to get both the portable GodPrompt Agent Skill and the MCP server in one step:

gemini extensions install https://github.com/AKzar1el/god-prompt-mcp

Or register only the published local stdio server with Gemini CLI's native MCP command:

gemini mcp add --scope user god-prompt npx -- -y god-prompt-mcp

The explicit user scope keeps this direct registration out of the project's .gemini/settings.json. Run gemini extensions list to verify the extension or gemini mcp list to verify the direct MCP registration. Node.js 22+ is required for the MCP server; GodPrompt itself needs no API key.

Google Antigravity CLI

Antigravity CLI supports Agent Plugins that bundle Agent Skills and MCP servers. Install the existing GodPrompt plugin directly from GitHub:

agy plugin install https://github.com/AKzar1el/god-prompt-mcp

Run agy plugin list to verify the installation. The repository's Agent Plugins 1.0 package supplies both skills/god-prompt and the pinned local stdio MCP server, so no separate skill or MCP registration is needed. Node.js 22+ is required for the MCP server.

GitHub CLI Agent Skills (preview)

GitHub CLI 2.100+ can install the same portable Agent Skill directly into supported coding agents without cloning the repository:

gh skill install AKzar1el/god-prompt-mcp skills/god-prompt --agent codex --scope user

Replace codex with another supported host such as github-copilot, claude-code, or cursor. The explicit skills/god-prompt path intentionally selects the portable skill rather than another distribution-specific package in this repository.

Vercel Skills CLI

The open skills CLI can install the same portable Agent Skill into Codex, Claude Code, Cursor, and many other supported coding agents:

npx -y skills@latest add AKzar1el/god-prompt-mcp --skill god-prompt --agent codex --copy -y

Replace codex with another supported agent when needed. The current CLI discovers the repository's portable god-prompt skill directly and installs SKILL.md plus its referenced protocol files; the MCP server remains separately available through npx -y god-prompt-mcp.

AllAgents

For a project that uses several coding agents, AllAgents can consume this repository once and synchronize GodPrompt to the selected clients. This verified example installs the portable skill for Claude Code, Codex, and Cursor while reusing the repository's bundled MCP configuration:

npx -y allagents@latest skill add AKzar1el/god-prompt-mcp --skill god-prompt --scope project --client claude,codex,cursor --yes

AllAgents resolves the repository as the source of truth, copies the god-prompt skill into each selected client, and materializes the existing local MCP server configuration with the package version already pinned by this repository. Use a different supported client set when your workspace requires it; no separate GodPrompt wrapper or package is needed.

Cline

Cline SDK / Hub 0.0.83+ can load GodPrompt's existing Agent Plugins 1.0 package directly. Cline discovers user-installed plugin packages under ~/.agents/plugins/*, validates the root plugin.json, exposes valid skills/ entries, and starts valid servers from the root mcp.json without writing them into cline_mcp_settings.json.

Install the repository into Cline's user plugin directory:

mkdir -p ~/.agents/plugins
git clone --depth 1 https://github.com/AKzar1el/god-prompt-mcp ~/.agents/plugins/god-prompt-mcp

On the next settings refresh or session build, Cline can expose the god-prompt Agent Skill and launch the bundled local stdio server, whose root mcp.json pins the published GodPrompt MCP package version. Cline intentionally does not auto-scan workspace .agents/plugins directories, so merely opening a repository cannot activate repository-controlled MCP servers; SDK hosts that deliberately trust another package root can opt in through agentPluginPaths.

Amp

Amp can load GodPrompt as an Agent Skill and expose the MCP tools only when that skill is relevant. Install the portable skill with:

npx -y skills@latest add AKzar1el/god-prompt-mcp --skill god-prompt --agent amp --copy -y

The skill includes Amp's sibling mcp.json configuration, pinned to the current GodPrompt MCP package. Amp keeps those MCP tool definitions hidden until the skill loads, avoiding a permanently expanded tool context while preserving the full progressive-disclosure surface. Node.js 22+ is required for the MCP server.

Factory Droid

Factory Droid can use GodPrompt as both an Agent Skill and a local MCP server. Install the complete portable skill with the current Skills CLI target for Droid:

npx -y skills@latest add AKzar1el/god-prompt-mcp --skill god-prompt --agent droid --copy -y

Droid discovers compatible project skills under .agents/skills/. To add the callable GodPrompt tools as a local stdio MCP server, use Droid's native MCP command with the published version pinned:

droid mcp add god-prompt "npx -y god-prompt-mcp@1.4.2"

Open /mcp in Droid to verify the server and available tools. The Agent Skill supplies the reusable workflow while MCP exposes GodPrompt's read-only task and protocol tools. Node.js 22+ is required for the MCP server.

Warp

Warp can use GodPrompt as both an Agent Skill and a local MCP server. Install the complete portable skill into Warp with:

npx -y skills@latest add AKzar1el/god-prompt-mcp --skill god-prompt --agent warp --copy -y

Warp discovers project skills from .agents/skills/ and .warp/skills/. For MCP tools, add a CLI-based MCP server in Warp with command npx and arguments -y, god-prompt-mcp; Warp's Oz CLI can also receive MCP configuration through oz agent run --mcp. The skill supplies the reusable workflow while MCP exposes GodPrompt's 22 public tools in v1.4.2. Node.js 22+ is required for the MCP server.

Goose

Goose can use GodPrompt as both an Agent Skill and a local MCP extension. Install the complete portable skill into Goose with:

npx -y skills@latest add AKzar1el/god-prompt-mcp --skill god-prompt --agent goose --copy -y

Goose discovers the copied skill under .goose/skills/god-prompt/. To expose GodPrompt's callable tools for a session, start Goose with the published MCP package as an external stdio extension:

goose session --with-extension "npx -y god-prompt-mcp@1.4.2"

Goose currently treats Agent Skills and MCP extensions as separate native surfaces, so this path does not depend on Agent Plugins support. The skill supplies the reusable workflow while MCP exposes GodPrompt's 22 public tools in v1.4.2. Node.js 22+ is required for the MCP server.

Zed

Zed can use GodPrompt as both a native Agent Skill and a local MCP context server. Install the complete portable skill, including its referenced protocol files, with:

npx -y skills@latest add AKzar1el/god-prompt-mcp --skill god-prompt --agent zed --copy -y

For MCP tools, open Settings → AI → MCP Servers → Add Server → Add Local Server, set the command to npx, and pass -y and god-prompt-mcp as arguments. Zed loads project skills from .agents/skills/ and can forward configured MCP servers to supported external agents. Node.js 22+ is required for the MCP server.

OpenCode

OpenCode v2 natively discovers Agent Skills from .agents/skills/ and supports local MCP servers. Install the same portable GodPrompt skill with:

npx -y skills@latest add AKzar1el/god-prompt-mcp --skill god-prompt --agent opencode --copy -y

To expose GodPrompt's MCP tools as well, register the published local server with OpenCode's native v2 command, then verify the connection:

opencode mcp add god-prompt -- npx -y god-prompt-mcp
opencode mcp list

The add command writes the project MCP configuration for you. If you prefer to configure it by hand, use opencode.json or opencode.jsonc:

{
  "$schema": "https://opencode.ai/config.json",
  "mcp": {
    "servers": {
      "god-prompt": {
        "type": "local",
        "command": ["npx", "-y", "god-prompt-mcp"]
      }
    }
  }
}

OpenCode 1.x used the same server entry directly under mcp; the v2 schema shown above nests named servers under mcp.servers.

The Agent Skill provides the reusable workflow; MCP adds the 22 public read-only tools available in v1.3.0. Node.js 22+ is required for the MCP server.

TraeCode

TraeCode can use GodPrompt as both a native project skill and a project-level local MCP server. Install the complete portable skill with the current Skills CLI target for Trae:

npx -y skills@latest add AKzar1el/god-prompt-mcp --skill god-prompt --agent trae --copy -y

The Trae target installs the skill under .trae/skills/god-prompt/, where TraeCode can load it when the task matches its description. To expose GodPrompt's callable tools for the project as well, add .trae/mcp.json:

{
  "mcpServers": {
    "god-prompt": {
      "command": "npx",
      "args": ["-y", "god-prompt-mcp@1.4.2"]
    }
  }
}

TraeCode also supports the portable .agents/skills/ convention when Enable .agents Skills Directory is turned on in its import settings. Treat project MCP configuration as executable workspace configuration and enable it only in repositories you trust. The Agent Skill supplies the reusable workflow while MCP exposes GodPrompt's 22 public tools in v1.4.2. Node.js 22+ is required for the MCP server.

OpenClaw

OpenClaw can use GodPrompt as both a project Agent Skill and a locally registered MCP server. Install the complete portable skill with the current Skills CLI target for OpenClaw:

npx -y skills@latest add AKzar1el/god-prompt-mcp --skill god-prompt --agent openclaw --copy -y

The copied skill lands under .agents/skills/god-prompt/, one of OpenClaw's documented project skill roots. To expose GodPrompt's callable tools to OpenClaw-managed agent runtimes as well, register the published stdio server:

openclaw mcp add god-prompt --command npx --arg -y --arg god-prompt-mcp@1.4.2
openclaw mcp probe god-prompt

mcp probe opens a live MCP connection and reports the discovered capabilities, making it a direct verification step after registration. The Agent Skill supplies the reusable workflow while MCP exposes GodPrompt's 22 public tools in v1.4.2. Node.js 22+ is required for the MCP server.

Qoder

Qoder can use GodPrompt as both a native project Skill and a project-level local MCP server. Install the portable skill with Qoder's supported Skills CLI target:

npx -y skills@latest add AKzar1el/god-prompt-mcp --skill god-prompt --agent qoder --copy -y

The Qoder target installs the skill under .qoder/skills/god-prompt/, where Qoder can load it automatically when the task matches the skill description. To expose GodPrompt's callable tools to the project as well, add .mcp.json at the project root:

{
  "mcpServers": {
    "god-prompt": {
      "command": "npx",
      "args": ["-y", "god-prompt-mcp@1.4.2"]
    }
  }
}

Qoder requires approval before using project-level MCP servers by default. After adding or changing the server, start a new session or run /mcp reload, then use /mcp to verify the connection. The Agent Skill supplies the reusable workflow while MCP exposes GodPrompt's 22 public tools in v1.4.2. Node.js 22+ is required for the MCP server.

Qwen Code

Qwen Code can use GodPrompt through both of its native extension surfaces: Agent Skills and local MCP servers. Install the complete portable skill with the current Skills CLI target for Qwen Code:

npx -y skills@latest add AKzar1el/god-prompt-mcp --skill god-prompt --agent qwen-code --copy -y

This installs GodPrompt under .qwen/skills/god-prompt/, where Qwen Code can discover and load the skill when the task matches its description. To add the callable GodPrompt tools as a project-local stdio MCP server, add this to .qwen/settings.json:

{
  "mcpServers": {
    "god-prompt": {
      "command": "npx",
      "args": ["-y", "god-prompt-mcp"]
    }
  }
}

Restart Qwen Code after changing MCP configuration, then open /mcp to verify the server. The Agent Skill provides the reusable workflow while MCP exposes GodPrompt's 22 public tools in v1.4.2. Node.js 22+ is required for the MCP server.

Kimi Code

Kimi Code CLI can install GodPrompt as one native custom plugin containing the existing Agent Skill and local MCP server. Install the current repository branch explicitly:

/plugins install https://github.com/AKzar1el/god-prompt-mcp/tree/main
/reload

The Kimi manifest points at skills/, so the god-prompt skill remains available through Kimi's normal relevance-based skill loading instead of being forced into every session. The same manifest exposes god-prompt-mcp@1.4.2 as a local stdio MCP server, including the 22 public tools available in v1.4.2. The explicit tree/main URL matters because Kimi's bare GitHub-repository install form prefers the latest GitHub release; using main makes the current plugin manifest available without forcing a documentation-only npm/MCP Registry release. Node.js 22+ is required for the MCP server.

JetBrains AI Assistant

JetBrains AI Assistant 2026.2 can use GodPrompt through both supported agent surfaces:

  • Agent Skill: in Settings → Tools → AI Assistant → Skills, open Manage External Registries and add https://github.com/AKzar1el/god-prompt-mcp. The existing skills/god-prompt package can then be installed for supported skill-aware agents such as Codex or Claude Agent.

  • MCP tools: in Settings → Tools → AI Assistant → Model Context Protocol (MCP), add a stdio server with command npx and arguments -y, god-prompt-mcp. Enable Pass custom MCP servers for the coding agent that should receive GodPrompt's progressive-disclosure tools.

The Agent Skill supplies the reusable workflow; the MCP server supplies GodPrompt's callable tools. Node.js 22+ is required for the MCP server.

JetBrains Junie CLI

Junie CLI can install GodPrompt as one extension by reusing this repository's existing Claude-compatible marketplace manifest. In Junie, register the repository and install the listed extension:

/extensions marketplace add AKzar1el/god-prompt-mcp
/extensions install god-prompt-mcp

Junie supports .claude-plugin/marketplace.json as a marketplace format, so no Junie-specific duplicate manifest is required. The installed extension reuses GodPrompt's existing portable Agent Skill and MCP configuration; Junie can also discover the same skill from .agents/skills/ and supports local npx MCP servers. Node.js 22+ is required for the MCP server. Junie CLI is not installed in this repository's qualification environment, so these commands are documented from JetBrains' current extension contract rather than claimed as a local end-to-end runtime smoke.

Devin Desktop / Windsurf

Devin Desktop (formerly Windsurf) and Devin CLI can load GodPrompt's existing Agent Plugins 1.0 package directly from GitHub. After signing in to Devin, install the plugin with:

devin plugins install AKzar1el/god-prompt-mcp

The plugin supplies both skills/god-prompt and the published local stdio MCP server. Run devin plugins list to verify the installation.

If you only want the portable skill, place the complete skill directory at .agents/skills/god-prompt/; Devin also discovers project skills under .devin/skills/ and .windsurf/skills/. If you only want the MCP server, register it directly:

devin mcp add god-prompt -- npx -y god-prompt-mcp

Run devin mcp list to verify the direct server registration. The Agent Skill provides the reusable workflow while MCP exposes GodPrompt's 22 public tools in v1.4.2. Node.js 22+ is required for the MCP server.

Cursor

The repository includes a native Cursor plugin manifest plus a portable Agent Plugins 1.0 manifest. Cursor Marketplace can use .cursor-plugin/plugin.json with cursor-mcp.json; Agent Plugins-compatible clients can use root plugin.json with mcp.json. Both launch the same local stdio server through npx -y god-prompt-mcp.

GitHub Copilot

GitHub Copilot CLI supports the portable Agent Plugins 1.0 package at the repository root. Install it through the repository marketplace:

copilot plugin marketplace add AKzar1el/god-prompt-mcp
copilot plugin install god-prompt-mcp@god-prompt

The plugin includes both the GodPrompt MCP server and a standards-compatible god-prompt Agent Skill under skills/god-prompt/. Copilot can load the skill automatically when relevant or invoke it explicitly as /god-prompt, while MCP remains available for progressive-disclosure tool access.

This keeps the marketplace entry, portable plugin, skill, and MCP server in one versioned repository and works for individual, team, and cloud-agent distribution. Direct repository installs (copilot plugin install AKzar1el/god-prompt-mcp) still work in current Copilot CLI releases, but the CLI marks them deprecated in favor of plugin@marketplace installs.

GitLab Duo CLI

GitLab Duo CLI 9.15+ supports Agent Plugins and Git-backed plugin marketplaces. Register this repository once, then install the same portable plugin:

duo plugin marketplace add https://github.com/AKzar1el/god-prompt-mcp.git
duo plugin install god-prompt-mcp@god-prompt

The installed plugin bundles the existing god-prompt Agent Skill and pinned local MCP server, so GitLab Duo users get both workflow guidance and callable tools from the same versioned repository.

Claude Code

The repository includes a Claude Code marketplace catalog alongside the existing plugin and MCP manifests, so users can install the same versioned plugin directly from this GitHub repository without waiting for community-directory review. In Claude Code, run:

/plugin marketplace add AKzar1el/god-prompt-mcp
/plugin install god-prompt-mcp@god-prompt

The plugin bundles the portable god-prompt Agent Skill and the pinned local MCP server. The same package remains structurally ready for Anthropic's reviewed community directory without maintaining a separate implementation.

Claude Code custom-gateway troubleshooting: In Claude Code 2.1.304+ with a third-party Anthropic-compatible gateway configured through ANTHROPIC_BASE_URL, an upstream tool-discovery regression can leave subagents without Read or headless claude -p without Agent/Bash. This is a Claude Code host issue, not a GodPrompt MCP server failure. Confirm the tools actually exposed in the affected session before applying a workaround.

Affected users report that merging "ENABLE_TOOL_SEARCH": "false" into the env section of their existing Claude Code settings restores core tools. This disables deferred tool discovery and can increase context-token usage; remove the workaround after an upstream fix. No change is needed for installations that already expose the expected tools.

Claude Desktop extension

GitHub releases include a .mcpb bundle for one-click local installation in MCPB-compatible clients such as Claude Desktop. The bundle runs the same local stdio server and does not require an API key or account.

Current GitHub releases are immutable and include a release attestation. With GitHub CLI installed, verify the release and your downloaded MCPB before installing it:

gh release verify <release-tag> --repo AKzar1el/god-prompt-mcp
gh release verify-asset <release-tag> <downloaded-mcpb> --repo AKzar1el/god-prompt-mcp

The asset check verifies that the local MCPB exactly matches the file recorded in the published release attestation.

Local stdio

npm install
npm run build
node dist/stdio.js

Example client configuration:

{
  "mcpServers": {
    "god-prompt": {
      "command": "node",
      "args": ["/absolute/path/to/god-prompt-mcp/dist/stdio.js"]
    }
  }
}

Privacy Policy

GodPrompt MCP can run either as a local stdio server or through the public Streamable HTTP Worker. Its content tools return static GodPrompt material bundled with the server, and classify_task is deterministic; neither mode calls a model or third-party content API.

  • Data collection and use: GodPrompt application code has no telemetry or analytics and does not request account data. Local stdio tool input stays in the local Node.js process. Hosted remote tool input is sent over HTTPS to the Cloudflare-hosted Worker so it can produce the MCP response.

  • Storage and retention: the MCP request handlers are stateless and have no application database or request-retention path; they do not intentionally persist tool inputs after a request completes. Hosting/provider infrastructure may process connection or request metadata under its own policies.

  • Third-party sharing: local stdio does not transmit tool inputs or bundled GodPrompt content to an external API. The hosted remote necessarily passes the request through Cloudflare infrastructure, but GodPrompt application code does not forward tool inputs to another API or service. The MCP host or AI client may process conversation and tool data under its own policies independently of this server.

  • Contact and policy: see tomiseregi.si/privacy for the current privacy policy and contact information. For support or bug reports, use GitHub Issues.

Development

npm install
npm run build
npm test

npm test builds the stdio server, performs the MCP initialization handshake, and verifies the expected tool list.

Repository discovery metadata is kept in server.json for MCP Registry-compatible consumers and glama.json for Glama.

Updating Content

To update the embedded GodPrompt content:

  1. Pull the latest content from the GodPrompt repository.

  2. Run npm run generate-content.

  3. Run npm test before publishing a new server release.

The generator reads the current source layout: GodPrompt.md, SKILL.md, and the three files under references/.

License

MIT

Project page: GodPrompt — AI software-development system prompt · GodPrompt source

Related MCP servers: Google Search Console · GEO Tracker · Web Validator · Google News & Trends

P4 shipped in v1.3.0

The five P4 diagnostic tools listed above and the paired evaluation enhancements shipped in the versioned v1.3.0 source, npm and MCP Registry release. The reproducible 74-case legacy fixture corpus was retained, with additional independent synthetic holdout cases. All evidence remains advisory and unverified unless separately authenticated by the host. See P4 release documentation.

Unpublished combined OpenAI Skill + MCP preview

The reviewed combined OpenAI plugin source is now in distribution/openai/ on main, while the published marketplace listing remains unchanged. It contains the existing scoped Skill plus a single remote MCP declaration for the current 22 public GodPrompt v1.3.0 tools. Two metadata formats (portable mcp.json and Codex compatibility .mcp.json) describe the same URL; they are not separate services. A dedicated GitHub workflow builds an unpublished review ZIP and validates the hosted tool list and a read-only call. The currently published OpenAI directory listing remains Skill-only until its separate reviewed update. See docs/qa/godprompt-openai-combined-preview-20261120.md.

P5 v1.4.2: The local repo mapper additionally recognizes test files that import built dist/*.js modules and maps them to matching existing src/*.ts source files. The mapping is advisory, not proof of test coverage. Local efficiency and regression measurements show measured MCP I/O and initial test-selection usefulness, not an independently verified coding-agent productivity gain.

P5 — local intelligence, benchmark and optional visualization

After installing the versioned Node 22+ npm distribution, the local-only repo mapper can generate a compact import/export graph without transmitting source code. For example, in a working directory you already own, run godprompt-repo-map C:/path/to/authorized-repo src/example.ts (change paths to match your checkout). The JSON output is heuristic: dynamic imports, aliases, symlink/hardlink relationships and test coverage still require review.

To collate verification receipts, use godprompt-proof pack --contract contract.json --repo OWNER/REPO --sha EXACT40CHAR_COMMIT --map checks.json, optionally with --snapshot connector-export.json and --codexify codexify-export.json. It reuses the existing evidence adapters, never executes a test or approves a release, and does not turn self-reported results into authenticated proof.

The paired benchmark generates deterministic matched-task order and scoring, and includes twelve independently failing microtask tests as initial grading fixtures. It does not run models or claim causal productivity gains. P5 design, security and limits.

MCP Apps-compatible clients can optionally present read-only, keyboard-accessible trajectory and impact reports through ui://godprompt/evidence-review.html. Text and structuredContent remain available to clients without MCP Apps and remain the basis for verification. No remote repository access or automatic policy execution has been added to the Worker.

Available Tools

22 tools
assess_release_readinessAssess reported release readinessA
Read-onlyIdempotent
Inspect

Check caller-reported CI checks, source/artifact revision, rollback, migration and approvals for missing blockers; never authorize deployment or authenticate receipts.

Inputs: checks (required), target (required), rollback_plan, migration_plan, +4 more.

audit_agent_instructionsAudit agent instruction consistencyA
Read-onlyIdempotent
Inspect

Review bounded caller-supplied instruction documents for duplicates, possible opposing action constraints and risky bypass wording. No instruction is executed or promoted.

Inputs: instructions (required). Returns: status, warning, findings, truncated, +2 more

audit_agent_trajectoryAudit reported agent trajectoryA
Read-onlyIdempotent
Inspect

Read-only review of bounded, sanitized event metadata for repeated failures, scope drift, stale verification and premature completion. No raw logs or independent execution evidence; cannot authorize an agent.

Inputs: events (required), budget_tokens. Returns: status, warning, findings, truncated, +8 more

audit_requirementsAudit software requirementsA
Read-onlyIdempotent
Inspect

Inspect bounded requirement text for missing acceptance scenarios, qualitative obligations and duplicates. Returns source-line-linked clarification prompts; cannot validate business correctness.

Inputs: specification (required).

build_test_matrixPlan requirement-traceable testsB
Read-onlyIdempotent
Inspect

Draft bounded positive, negative, boundary and regression test cases for supplied requirements. Reuses the requirements parser, reports unspecified expectations and never claims tests ran.

Inputs: specification (required), max_cases.

classify_taskClassify software-development taskA
Read-onlyIdempotent
Inspect

Classify one concrete software-development task into one of GodPrompt's 9 task types (BUILD, DEBUG, REFACTOR, CONTENT, DESIGN, SHIP, ANALYZE, AUTOMATE, PLAN), or UNCLASSIFIED when no reliable route is detected. Returns JSON with task_type, deterministic confidence, protocol, matched_signals, alternative_task_types, ambiguous, and recommendation; use it for routing before loading detailed workflow text.

Inputs: description (required).

compare_evaluation_runsCompare paired agent evaluation runsA
Read-onlyIdempotent
Inspect

Compare bounded caller-reported outcomes on exact matching case IDs, with coverage exclusions and low-sample warnings. No API calls, statistical significance claims or stored results.

Inputs: baseline (required), candidate (required).

create_context_handoffPrepare a safe cross-chat task handoffB
Read-onlyIdempotent
Inspect

Compose a bounded, explicitly unverified state transfer from caller-reported decisions, open items and evidence references, with best-effort token/email redaction.

Inputs: goal (required), evidence, completed, decisions, +1 more.

create_work_contractCreate a bounded GodPrompt task contractA
Read-onlyIdempotent
Inspect

Create a portable, SHA-256-fingerprinted task contract with explicitly provided scope, acceptance criteria, verification commands and prohibitions. Deterministic, read-only and stateless. Use classify_task first; missing details produce NEEDS_DEFINITION, never inferred commands or permission to execute.

Inputs: task (required), task_type, risk_level, allowed_paths, +3 more.

design_adversarial_testsDesign adversarial and property test casesA
Read-onlyIdempotent
Inspect

Generate bounded negative, boundary, metamorphic, property and trust-boundary test ideas from explicit caller-supplied field contracts. Never runs or generates executable tests.

Inputs: fields (required), max_cases. Returns: cases, issues, status, warning, +4 more

diagnose_failureDiagnose supplied failure symptomsA
Read-onlyIdempotent
Inspect

Return bounded evidence-class hypotheses and differentiating experiments for supplied symptoms/logs. Does not expose raw logs or claim a root cause.

Inputs: symptoms (required), logs.

get_anti_patternsGet GodPrompt anti-patternsA
Read-onlyIdempotent
Inspect

Returns references/03-ANTI-PATTERNS.md — red flags, rationalizations, and recovery patterns. Covers the 10 most dangerous anti-patterns that lead to broken code, scope creep, and false confidence. Load when you catch yourself rationalizing. ~9KB.

Inputs: no named parameters.

get_core_skillGet core GodPrompt skillA
Read-onlyIdempotent
Inspect

Returns SKILL.md — the lean core protocol (~32KB) covering the universal 6-phase protocol, Three Iron Laws, and task auto-classification. Load it at the start of a task or after a context reset, then reuse that context instead of reloading it on every message. Start here for progressive disclosure.

Inputs: no named parameters.

get_gatesGet GodPrompt verification gatesA
Read-onlyIdempotent
Inspect

Returns references/02-GATES.md — verification checklists, THE GATE (pre-completion verification), and structured report templates for every deliverable type. Load when you need to verify work before claiming completion. ~9KB.

Inputs: no named parameters.

get_god_promptGet full GodPromptA
Read-onlyIdempotent
Inspect

Returns the complete GodPrompt.md — a single-file universal system prompt for AI software development (63KB, ~1145 lines). With no arguments, returns the original full text for backward compatibility. Bridges may truncate the middle of such a large response: pass start (initially 0) and maxChars (at most 8192) to retrieve bounded, lossless chunks; structuredContent.nextStart gives the next offset, or null at EOF. For progressive disclosure, use get_core_skill and references instead.

Inputs: start, maxChars.

get_protocolsGet GodPrompt protocolsA
Read-onlyIdempotent
Inspect

Returns references/01-PROTOCOLS.md — deep execution guides for each task type (BUILD, DEBUG, REFACTOR, CONTENT, DESIGN, SHIP, ANALYZE, AUTOMATE, PLAN). Load this when the task requires detailed protocol steps beyond the core skill. ~13KB.

Inputs: no named parameters.

get_tailored_guidanceGet phase-specific GodPrompt guidanceB
Read-onlyIdempotent
Inspect

Return only the relevant task-type protocol or quality gate for one phase, paged into small bounded chunks. More efficient than loading the entire reference; no tool execution or verification.

Inputs: task_type (required), phase, start, max_chars.

get_versionGet GodPrompt versionsA
Read-onlyIdempotent
Inspect

Returns JSON with the GodPrompt content version, MCP server version, repository, bundled file sizes and purposes, and task-type/protocol catalog. Use it to verify which GodPrompt content/server version a client is connected to; use the content retrieval tools when you need the actual workflow text.

Inputs: no named parameters.

review_code_changeReview supplied code changesA
Read-onlyIdempotent
Inspect

Analyze a bounded unified diff for line-anchored risk signals, changed-file coverage, and review follow-ups. Heuristics only: not a semantic review, vulnerability scanner or verified test result. Stateless, no repository access.

Inputs: diff (required).

review_mcp_tool_contractsReview supplied MCP contractsA
Read-onlyIdempotent
Inspect

Read-only bounded design audit of MCP tool descriptions, input/output schemas and annotations. Not a protocol validator, vulnerability scanner or permission grant.

Inputs: tools (required). Returns: issues, status, warning, truncated, +2 more

review_work_evidenceReview task completion evidenceA
Read-onlyIdempotent
Inspect

Fail-closed audit of a create_work_contract result against supplied changed paths, command receipts, criteria, prohibited actions and risk review. Returns HOLD or REPORTED_READY_FOR_INDEPENDENT_REVIEW; never independently verifies caller-supplied evidence or authorizes completion, merge or deployment.

Inputs: contract (required), checks, acceptance, observations, +3 more.

trace_change_impactTrace changed-code impact and test coverageA
Read-onlyIdempotent
Inspect

Rank affected files and select test cases with explicit dependency and requirement-to-test mappings. All graph and receipt evidence must be supplied by caller; no repository or command execution.

Inputs: changed_files (required), tests, receipts, dependencies, +1 more. Returns: issues, status, warning, truncated, +6 more

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelog…

    TDQS

    A3.7/5.0

    Scored across 22 tools

    Disambiguation4/5

    Most tools target a distinct artifact and action (classify_task, review_code_change, create_work_contract), but progressive-disclosure retrieval tools such as get_tailored_guidance versus get_protocols/get_gates overlap in purpose. The two test-generation tools, build_test_matrix and design_adversarial_tests, also have adjacent boundaries, though descriptions mostly resolve the distinction.

    Naming Consistency5/5

    All names use consistent snake_case action_object phrasing, with clear verb families such as get_, review_, audit_, create_, and build_. There is no mixed casing or vague one-word naming.

    Tool Count3/5

    At 22 tools, the server sits in the heavy 16-25 range and requires agents to scan a large surface before choosing. Each tool appears to earn a place in a broad GodPrompt workflow suite, but the count is borderline for easy navigation.

    Completeness4/5

    The surface covers content retrieval, task classification, contract creation, multiple audit/review dimensions, test ideation, release readiness, handoff, and evidence review. Remaining gaps, such as no execution/verification, no persistence, and no semantic review, appear to be intentional scope limits rather than accidental dead ends.

    Maintenance

    ActivityActive
    ResponsivenessNo recent community issues

    Related MCP Connectors

    Related MCP Servers

    • F
      license
      Not graded
      quality
      D
      maintenance
      Provides AI-powered selection and generation of specialized system prompts from a database of over 66 templates for Claude Code. It uses semantic search to find the best matching template and can adapt it to fit specific user tasks and contexts.
      -
    • A
      license
      A
      quality
      D
      maintenance
      Automatically generates and manages a prompts system for software projects, enabling persistent context for AI coding assistants through project scanning, requirement clarification, and module tracking.
      8
      MIT
    • A
      license
      A
      quality
      B
      maintenance
      Enables refining prompt intent, structure, constraints, and output instructions before prompts reach an LLM, with tools for schema retrieval, brief validation, and prompt variant generation.
      3
      MIT