Overview
RunTools supports Model Context Protocol (MCP) - an open standard for connecting AI models to data sources and tools.Adding MCP Tools
Include MCP servers in your agent runtime:MCP Server Types
NPX Servers
Run MCP servers via npx:Docker Servers
Run MCP servers in Docker:Remote Servers
Connect to remote MCP servers:Available MCP Servers
Official (Anthropic)
| Server | Description |
|---|---|
@anthropic/mcp-server-filesystem | File system access |
@anthropic/mcp-server-github | GitHub integration |
@anthropic/mcp-server-slack | Slack integration |
@anthropic/mcp-server-postgres | PostgreSQL queries |
Community
Many community MCP servers are available on npm and GitHub.Configuration
Environment Variables
Inject secrets into MCP servers:Arguments
Pass arguments to MCP servers:Tool Discovery
MCP servers expose their tools automatically. The agent discovers available tools at runtime.Best Practices
Use official servers when possible
Use official servers when possible
Official Anthropic servers are well-maintained and secure.
Pin versions
Pin versions
Use specific versions:
@anthropic/[email protected]Secure credentials
Secure credentials
Use RunTools secrets for MCP server credentials.
Limit server scope
Limit server scope
Give MCP servers only the access they need.
Creating MCP Servers
Build your own MCP servers:MCP Documentation
Learn more about Model Context Protocol