mcp-farmer
A CLI tool for scaffolding, testing, extending and analyzing MCP (Model Context Protocol) servers
claude mcp add --transport stdio boldare-mcp-farmer npx -y mcp-farmer
How to use
mcp-farmer is a comprehensive CLI for working with MCP servers. It lets you vet, document, and explore MCP servers, as well as generate MCP tools from API specs, and probe or grow existing servers. Core commands include vet (audit and generate reports), doc (generate HTML documentation), new (scaffold a new MCP server), market (browse popular MCP servers), try (interactively call tools/resources/prompts on a server), grow (generate tools from OpenAPI/GraphQL specs), probe (test tools with AI-generated inputs). Use these tools to assess server quality, quickly bootstrap tooling, and produce shareable reports and documentation. The CLI supports both HTTP transports and stdio mode, enabling you to connect to a remote MCP server or spawn a local server process for testing. Commands can be combined with options to customize output formats and authentication flows (e.g., --oauth for HTTP servers).
How to install
Prerequisites:
- Node.js v20 or newer installed on your system (the project description recommends Node.js >= 20).
- Internet access to install packages from npm/yarn/bun as needed.
Install the MCP farmer CLI globally (recommended):
npm install -g mcp-farmer
Alternative package managers (examples):
bun install -g mcp-farmer
Quick test without a global install:
npx mcp-farmer try http://localhost:3000/mcp
If you prefer to use local/project-scoped usage, you can also run via npx:
npx -y mcp-farmer try http://localhost:3000/mcp
Basic workflow example:
- Vet a server and export a report:
mcp-farmer vet http://localhost:3000/mcp --output markdown > report.md
- Generate documentation for a local server example:
mcp-farmer doc --local "npx -y @example/mcp-server" --out docs.html
Click-through help for all commands:
mcp-farmer --help
Additional notes
Notes and tips:
- mcp-farmer can auto-detect MCP servers from local configurations (Cursor, VS Code, Claude Desktop, Claude Code, OpenCode, Gemini CLI) during vet/probe workflows.
- For remote servers requiring authentication, you can use the --oauth flag to initiate an OAuth flow during vet or probe.
- When generating documentation with doc, you can specify --remote or --local installation methods, and use --env to document required environment variables for local servers or --header for remote servers' required headers. Use --out to write results directly to a file.
- The grow command generates MCP tools from API specs (OpenAPI or GraphQL) and requires an ACP agent (OpenCode, Claude Code, or Gemini CLI) to be installed to perform code generation.
- Vet reports can be exported in json/html/markdown formats; use -o to select the output format. Auto-detection may prompt you to choose among multiple servers found in config files.
- If you’re unsure of a server’s transport, recall that MCP servers can be reached via HTTP URL or via stdio (spawned process) after --. Some workflows demonstrate using npx and bunx for stdio-based tests.
Related MCP Servers
mcp-svelte-docs
🔍 MCP server that lets you search and access Svelte documentation with built-in caching
neurolink
Universal AI Development Platform with MCP server integration, multi-provider support, and professional CLI. Build, test, and deploy AI applications with multiple ai providers.
mcp-jest
Automated testing for Model Context Protocol servers. Ship MCP Servers with confidence.
local -gateway
Aggregate multiple MCP servers into a single endpoint with web UI, OAuth 2.1, and profile-based tool management
openapi
Powerful bridge between OpenAPI specifications and AI assistants using the Model Context Protocol (MCP). Automatically converts any OpenAPI/Swagger API specification into MCP tools that can be used by AI assistants like Claude Desktop.
mcpplay
The FastAPI /docs experience, for MCP servers.