Guide and evaluation harness for building high-quality MCP (Model Context Protocol) servers in TypeScript or Python.
Run the MCP evaluation harness against a user's MCP server using a QA-pair XML file. Wraps scripts/evaluation.py (stdio/http/sse transports) and returns a scored markdown report. Invoke with `/mcp-builder:evaluate <eval.xml> <server-description>` — server-description can be natural language describing how to launch or reach the server.
Use when the user wants to build, create, write, design, or scaffold an MCP (Model Context Protocol) server — triggers on phrases like "build an MCP server", "create an MCP server for X", "MCP server for {service}", "write tools for an LLM to call {API}", or "wrap {service} as an MCP server". Provides Anthropic's 4-phase workflow (research → implement → test → evaluate) with per-language references for TypeScript (recommended) and Python. Prefer this skill over generic MCP advice whenever the task involves authoring a new server.
A Claude Code plugin that packages Anthropic's guide for building high-quality MCP (Model Context Protocol) servers, plus an evaluation harness for scoring them against QA-pair test suites.
mcp-builder skill — Anthropic's 4-phase workflow (research → implement → test → evaluate) with per-language references for TypeScript (recommended) and Python. Activates automatically on prompts like "build an MCP server for X", "wrap the Y API as MCP tools", or "write an MCP server in TypeScript"./mcp-builder:evaluate slash command — Wraps scripts/evaluation.py. Takes a QA XML file plus a server description (stdio command, http URL, or sse URL) and returns a scored markdown report.From the Make Heavy Metal marketplace (recommended):
/plugin marketplace add make-heavy-metal/mcp-builder
/plugin install mcp-builder@make-heavy-metal
/plugin marketplace update make-heavy-metal pulls new versions when they ship.
For local development on the plugin itself: clone the repo and point Claude Code at it as a local marketplace:
git clone https://github.com/make-heavy-metal/mcp-builder.git
# then, inside Claude Code:
# /plugin marketplace add ./mcp-builder
# /plugin install mcp-builder@make-heavy-metal
For the evaluation harness only:
pip install -r scripts/requirements.txt # anthropic>=0.39.0, mcp>=1.1.0
export ANTHROPIC_API_KEY=...
The skill itself has no runtime dependencies — it's a guide the agent reads.
Build an MCP server — just describe what you want. The skill auto-activates:
"Build an MCP server in TypeScript that exposes the Linear API"
Evaluate a server — invoke the slash command with an eval file and a server description:
/mcp-builder:evaluate my-eval.xml stdio: node dist/index.js
/mcp-builder:evaluate my-eval.xml http://localhost:3000/mcp with header Authorization: Bearer $TOKEN
See scripts/example_evaluation.xml for the QA-pair format and skills/mcp-builder/references/evaluation.md for how to author good eval questions.
See LICENSE. Original content authored by Anthropic; packaged as a plugin by Travis Corrigan.
Own this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimOwn this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimBased on adoption, maintenance, documentation, and repository signals. Not a security audit or endorsement.
npx claudepluginhub make-heavy-metal/mcp-builder --plugin mcp-builderTools to maintain and improve CLAUDE.md files - audit quality, capture session learnings, and keep project memory current.
A growing collection of Claude-compatible academic workflow bundles. Covers scientific figures, manuscript writing and polishing, reviewer assessment, citation retrieval, data availability, paper reading, literature search, response letters, paper-to-PPTX conversion, and evidence-grounded Chinese invention patent drafting. Rules are organized as reusable skill folders with explicit workflows and quality checks.
Browser automation and end-to-end testing MCP server by Microsoft. Enables Claude to interact with web pages, take screenshots, fill forms, click elements, and perform automated browser testing workflows.
Comprehensive feature development workflow with specialized agents for codebase exploration, architecture design, and quality review
Core skills library for Claude Code: TDD, debugging, collaboration patterns, and proven techniques
Harness-native ECC operator layer - 67 agents, 271 skills, 92 legacy command shims, reusable hooks, rules, selective install profiles, and production-ready workflows for Claude Code, Codex, OpenCode, Cursor, and related agent harnesses