Skills Mcp

Name: Skills Mcp
Author: Jignesh-Ponamwar

@Jignesh-Ponamwar

Skills Mcp について

A semantic skills registry over MCP that teaches AI agents how to do things correctly using curated expert skills that agents discover and load at the moment they need them. You can build your own.

#agent-skills #skills #mcp

基本情報

カテゴリ

その他

トランスポート

stdio

公開者

Jignesh-Ponamwar

投稿者

Jignesh Ponamwar

設定

以下の設定を使って、このサーバーを MCP 対応クライアントに追加してください。

{
  "mcpServers": {
    "skill-mcp": {
      "transport": "streamable-http",
      "url": "https://skill-mcp.jigneshponamwar.workers.dev/mcp"
    }
  }
}

ツール

STEP 1 — Discover relevant skills. Call this FIRST at the start of any task to check whether the registry contains a curated skill that matches. Performs semantic vector search and returns ranked results with similarity scores. Workflow after this call: • score > 0.6 → strong match — call skills_get_body with that skill_id • score 0.4–0.6 → possible match — inspect description before proceeding • score < 0.4 → no relevant skill — proceed without one Query tips: be task-specific, not generic. 'write pytest unit tests for a Flask REST API' outperforms 'testing'. Describe what you are trying to accomplish, not what you want to find.

STEP 2 — Load full skill instructions. Call after skills_find_relevant once you have identified the best-matching skill_id. Returns three fields: • instructions — expert step-by-step guidance; read and follow these • system_prompt_addition — optional context to add to your persona (may be empty) • tier3_manifest — lists available references, scripts, and assets by filename After loading: apply the instructions. If tier3_manifest lists files that the instructions explicitly reference, fetch them with skills_get_reference, skills_run_script, or skills_get_asset. Most tasks are fully served by the instructions alone — do not load Tier 3 speculatively. Version pinning: pass version='1.2' to pin to a specific skill version, or use the inline form skill_id='[email protected]'. If the requested version is not found, the latest version is returned with a version_note explaining the fallback. Deprecated skills include a deprecation_notice field naming the replacement.

OPTIONAL STEP 2b — Load config schema, variants, and constraints for a skill. Call only when: (a) the user asks to customize skill behaviour, or (b) skills_get_body instructions mention configurable options. Returns: config_schema (JSON Schema for parameters), variants (alternative skill modes), dependencies (required tools/packages), limitations (known constraints). Do NOT call this by default — most tasks complete with skills_get_body alone.

STEP 3a — Fetch a reference document bundled with a skill (markdown files: checklists, policies, API specs, examples). Two-phase use: 1. Call with filename='list' (default) to see the full reference manifest 2. Call again with the specific filename to fetch its content Only call when: tier3_manifest from skills_get_body lists reference files AND the skill instructions explicitly name one. Do not load references speculatively.

STEP 3b — Execute a helper script bundled with a skill. Script source is NEVER returned — only stdout, stderr, and exit_code. Two-phase use: 1. Call with filename='list' to see available scripts and their descriptions 2. Call with the specific filename (and optional input_data) to execute input_data: key-value pairs passed to the script as environment variables. Deployment note: script execution requires the local server (MCP_TRANSPORT=streamable-http python -m skill_mcp.server). The Cloudflare Workers deployment returns the manifest only — Pyodide has no subprocess support. Only call when skill instructions direct you to run a specific script.

STEP 3c — Fetch a template or static resource bundled with a skill (markdown templates, config starters, example data files). Two-phase use: 1. Call with filename='list' (default) to see the full asset manifest 2. Call again with the specific filename to fetch its content Use the returned content as a starting template — adapt it to the specific task. Only call when skill instructions reference a specific asset file.

概要

What is Skills Mcp?

Skills Mcp is a self-hostable registry of expert AI agent skills delivered over MCP (Model Context Protocol). It provides AI agents with a searchable library of curated, step-by-step procedures (skills) for tasks like setting up Stripe webhooks, containerizing Python apps, or writing structured code reviews.

How to use Skills Mcp?

An agent calls skills_find_relevant with a natural language query, receives ranked matches with similarity scores, then loads full instructions via skills_get_body. Self-hosting requires cloning the repository and deploying to Cloudflare Workers with a free Qdrant Cloud cluster and Cloudflare Workers AI credentials; a Docker path is also available for fully local use.

Key features of Skills Mcp

Semantic vector search using Cloudflare Workers AI (bge-small-en-v1.5)
Three-tier progressive disclosure: discover, load body, fetch supplementary files
30 bundled skills covering Stripe, Supabase, FastAPI, Docker, and more
Self-hostable in under 10 minutes with no separate backend or GPU
Backed by Qdrant Cloud for vector storage
Only descriptions and trigger phrases are embedded; full instructions fetched on demand

Use cases of Skills Mcp

AI agent retrieving expert runbooks for software development tasks
Automating Stripe subscription setup with webhooks using verified procedures
Containerizing a Python application with step-by-step security checklists
Writing structured code reviews with common mistake avoidance
Generating PDF, Excel, Word, or PowerPoint documents via scripted skills

FAQ from Skills Mcp

What is the core idea behind Skills Mcp?

Instead of hoping an AI agent remembers the right way to perform a task, it gives the agent a searchable library of curated, step-by-step procedures it can pull from on demand.

How does skill discovery work?

Discovery uses semantic vector search (Cloudflare Workers AI) so natural language queries match correctly even when phrasing varies. Only the short description and trigger phrases are embedded; full instructions are fetched separately on demand.

What are the runtime dependencies?

You need a free Cloudflare Workers account, a free Qdrant Cloud cluster, and Cloudflare Workers AI credentials for embeddings. A Docker path is also available for fully local use without cloud services.

What skills are included?

30 bundled skills covering Stripe, Supabase, FastAPI, React, Next.js, Docker, GitHub Actions, Terraform, Cloudflare Workers, Anthropic/OpenAI/Gemini APIs, PDF/Excel/Word/PowerPoint generation, SQL, web scraping, code review, test writing, and MCP server building.

How is the server architected?

It runs as a single Cloudflare Python Worker backed by Qdrant Cloud. A seed script populates six Qdrant collections from SKILL.md files. Nothing is loaded speculatively; agents follow a three-tier progressive disclosure pattern.

「その他」の他のコンテンツ

MCP Toolbox for Databases

googleapis

MCP Toolbox for Databases is an open source MCP server for databases.

15.9K

Unity MCP ✨

justinpbarnett

Unity MCP acts as a bridge between AI assistants and your Unity Editor. Give your LLM tools to manage assets, control scenes, edit scripts, and automate tasks within Unity.

12.2K