MCP Pickaxe Server

# CLAUDE.md This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository. ## Project Overview MCP server for the Pickaxe API - provides AI assistants with tools to manage Pickaxe studios, AI agents, documents, users, and analytics. ## Commands ```bash npm install # Install dependencies npm run build # Compile TypeScript to dist/ npm run dev # Run in development mode (tsx, no build required) npm start # Run compiled version from dist/ ``` ## Architecture Single-file MCP server (`src/index.ts`) using `@modelcontextprotocol/sdk`: - **Multi-studio support**: Environment variables `PICKAXE_STUDIO_<NAME>` configure API keys for different studios. All tools accept an optional `studio` parameter. - **API wrapper**: `pickaxeRequest()` handles all Pickaxe API calls with auth and error handling - **Tool definitions**: Array of `Tool` objects with JSON Schema input definitions - **Tool execution**: `executeTool()` switch statement routes tool calls to API endpoints ## Multi-Studio Configuration Studios are configured via environment variables: ``` PICKAXE_STUDIO_PRODUCTION=studio-xxx PICKAXE_STUDIO_STAGING=studio-yyy PICKAXE_DEFAULT_STUDIO=PRODUCTION ``` The `getConfiguredStudios()` function scans env vars for `PICKAXE_STUDIO_*` prefixes. If no studio is specified in a tool call and no default is set, the server throws an error listing available options. ## Adding New Tools 1. Add tool definition to the `tools` array with name, description, and inputSchema 2. Add case handler in `executeTool()` switch statement 3. Use `pickaxeRequest(endpoint, method, body, studio)` to call the API ## Pickaxe API Base URL: `https://api.pickaxe.co/v1` Main endpoints used: - `/studio/pickaxe/history` - Agent chat logs - `/studio/document/*` - Knowledge base documents - `/studio/user/*` - User management - `/studio/product/list` - Products/bundles - `/studio/memory/*` - Memory schemas and user memories