mcp-pickaxe

MCP Pickaxe is an MCP server that connects AI assistants (like Claude) to the Pickaxe platform, enabling you to manage agents, knowledge bases, users, and analytics directly from natural language. It exposes a suite of tools for working with studios, chat history, documents, users, products, and memories, allowing you to analyze conversations, maintain knowledge bases, manage access, and monitor activity across multiple Pickaxe studios. Typical use cases include security monitoring workflows that scan chat history for prompt-injection patterns, auto-research pipelines that fact-check knowledge base articles, and cross-studio operations that let you switch contexts without manually swapping API keys. Tools available through the MCP server include: chat_history, studios_list, doc_list, doc_get, doc_create, doc_connect, user_list, memory_list, memory_get_user, products_list, and more. You can configure multiple studios by supplying environment variables and selecting a default studio when making requests. The server is designed to work with Claude Desktop and Claude Code configurations, enabling seamless integration into existing AI workflows.

Prerequisites:

• Node.js 18+ (LTS recommended)
• npm (bundled with Node.js) or pnpm/yarn as you prefer
• A Pickaxe account with API access and a Studio API key (starts with studio-)

Option 1: Install from npm (recommended)

1. Install the MCP server globally: npm install -g mcp-pickaxe
2. Run the server (example): npx mcp-pickaxe
3. If you need to specify a custom path or environment, use your preferred runner: npx mcp-pickaxe

Option 2: Clone and Build

1. Clone the repository: git clone https://github.com/aplaceforallmystuff/mcp-pickaxe.git
2. Install dependencies: cd mcp-pickaxe npm install
3. Build the project: npm run build
4. Run the built server (path may vary based on build output): node dist/index.js

Note: The README examples show typical usage patterns and configuration steps for Claude integrations. Adjust paths and keys to your environment.

Configuration tips:

• The MCP server expects a node-based runtime and a path to the built entry (dist/index.js). Update the mcp_config accordingly with the correct path after building.
• For multi-studio setups, you can provide multiple PICKAXE_STUDIO_... environment variables and a PICKAXE_DEFAULT_STUDIO to specify the default studio. When sending requests, you can override the studio in each call as needed.
• Environment variable to set the API key in Claude configurations: PICKAXE_STUDIO_MAIN (value should be your studio API key).
• If you encounter connectivity issues, ensure your Studio API key has the necessary scope and that network access to Pickaxe endpoints is allowed from your runtime environment.
• The npm package name is mcp-pickaxe; use the npm_package field to surface this in tooling that reads the manifest.