mcp-builder
MCP Server Development Tool
Build high-quality Model Context Protocol servers to provide external tool capabilities for AI agents.
Core Capabilities
- MCP protocol understanding and server architecture design
- Python/TypeScript SDK implementation
- Tool schema design for AI agents
- Evaluation test creation and validation
Tech Stack
| Language | SDK | Validation | Runtime |
|---|---|---|---|
| Python | MCP Python SDK | Pydantic v2 | asyncio |
| TypeScript | MCP TypeScript SDK | Zod | Node.js 18+ |
Executable Tools
The following scripts can be run directly without reading source code:
scripts/evaluation.py- Run MCP server evaluation testsscripts/connections.py- Test server connection status
Design Principles
- Workflow-Oriented: Build complete task tools, not simple API wrappers
- Optimize Context: Return high-signal information, avoid data dumping
- Actionable Errors: Error messages guide agents to correct usage
- Evaluation-Driven: Create evaluation scenarios early, iterate based on agent feedback
Boundaries
Focus on MCP server development and tool design, not third-party API development or client integration.
Detailed References
./workflows/mcp-development.md- Complete development workflow./guides/mcp_best_practices.md- MCP best practices./guides/python_mcp_server.md- Python implementation guide./guides/node_mcp_server.md- TypeScript implementation guide./guides/evaluation.md- Evaluation creation guide
More from jochenyang/jochen-ai-rules
ui-ux-pro-max
UI/UX design-system reasoning and UX quality audit skill. Use when user needs style direction, palette/typography selection, UX review, or design optimization before implementation. Do NOT use for backend logic or database design.
24devops-engineer
CI/CD pipeline design, containerization, and infrastructure management. Handles Docker, Kubernetes, monitoring setup (Prometheus/Grafana), and infrastructure-as-code (Terraform/Pulumi). Use when user asks to deploy, configure CI/CD, set up Docker/K8s, or manage infrastructure.
19reflect
Review current conversation, analyze tasks, errors, and user feedback, extract learning opportunities for skill improvement. Use when user says "reflect", "review session", "what did we learn", "session summary", or after completing a complex task.
1claude-audit
Audit all .claude/*.md files for redundant instructions, verbose phrasing, and content that could move to memory. Use when user wants to optimize their Claude configuration rules.
1miloya-codebase
Project context engine for repo orientation, cached handoff, and task-focused code retrieval.
1phaser-build
Build 2D HTML5 games for the web using Phaser with WebGL/Canvas rendering. Use when creating Phaser game scaffolds, scenes, core loops, or deployment guidance for web-based releases.
1