Mcp builder
Skill BAKUGOS1/SkilledAgents-Toolkit/plugins/skilled-agents/skills/mcp-builder
MCP server design, implementation, and review workflow for Model Context Protocol integrations. Use when building or auditing MCP servers, tools, resources, prompts, connectors, external API integrations, tool schemas, authentication boundaries, error handling, pagination, or agent-facing workflow tools in Python, TypeScript, or another runtime.From its SKILL.md
npx -y skills add BAKUGOS1/SkilledAgents-Toolkit --skill mcp-builderAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
One thing to look at
- 0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
SKILL.md
2.2 KB, 369 tokens by cl100k_base, as published. Nobody here has run it
MCP Builder
Overview
Use this skill to build MCP servers that are useful to agents in real workflows, not just thin API wrappers. It is adapted from useful patterns in Anthropic's Apache-2.0 mcp-builder skill.
First Principles
- Verify current MCP docs before relying on protocol details, SDK APIs, or transport behavior.
- Design tools around tasks agents need to complete.
- Prefer clear, discoverable tool names with consistent prefixes.
- Return focused data with pagination/filtering instead of huge responses.
- Make errors actionable: tell the agent what failed, why, and what to try next.
- Keep auth, secrets, destructive actions, and production writes behind explicit safeguards.
Design Workflow
- Identify the external service, target users, and top workflows.
- List required read tools, write tools, resources, and prompts.
- Define tool schemas with strict inputs, clear descriptions, and safe defaults.
- Decide which operations need confirmation or dry-run behavior.
- Implement the smallest useful server using the project's existing runtime.
- Add tests or smoke checks for connection, happy path, invalid input, auth failure, pagination, and rate-limit behavior.
- Document setup, environment variables, and example prompts.
Tool Quality Checklist
- Names are action-oriented and easy to search.
- Descriptions explain when to use the tool and what it returns.
- Inputs are typed, minimal, and validated.
- Outputs are structured and not overlong.
- Errors include next steps.
- Secrets never appear in logs or responses.
- Destructive tools require explicit confirmation.
- The server can be run locally and verified with a simple client.
Final Handoff
Report tools/resources added, auth/env requirements, verification performed, remaining risks, and any MCP docs consulted.
What ships with it: 1 file
165 B alongside SKILL.md
agents/
- openai.yaml165 B