| name | mcp-reference-servers |
| description | Guidance on using and integrating official reference servers. Use when building custom servers or deploying standard configurations. For server listings, see mcp-server-registry. |
MCP Reference Servers Skill
When to Use
- When building new MCP servers and needing a reference implementation.
- When the user wants robust, community-tested servers for common tools (Filesystem, Google Search, Fetch, etc.).
- When debugging MCP connection issues using standardized servers.
Reference
MCP Reference Servers GitHub
Instructions
1. Identify Target Server
- Review the
src directory in the reference repository to see available servers.
- Common servers include:
fetch: For web page scraping.
filesystem: For local file access.
google-search: For web searching capabilities.
postgres: For database interaction.
2. Implementation & Setup
- Most reference servers are built with Node.js/TypeScript or Python.
- Follow the
README.md within each sub-directory for build and run instructions.
- Ensure the appropriate runtime (Node.js 18+ or Python 3.10+) is installed.
3. Usage via MCP Configuration
{
"mcpServers": {
"server-name": {
"command": "npx",
"args": ["-y", "@modelcontextprotocol/server-name"],
"env": {
"ENV_VAR": "value"
}
}
}
}
Quality Gates
- Use
npx for quick access without local installation unless performance is a concern.
- Keep environment variables secure in protected configuration files.
- Verify the server's capabilities via
list_tools command after initialization.