float-mcp

# MCP Foundations & Key Concepts ## What is MCP? MCP (Model Context Protocol) is a protocol and server architecture for exposing structured, tool-based APIs to AI agents and clients. --- ## Core Concepts - **Tool**: A discrete, self-describing API operation (e.g., `list-people`, `create-project`). - **Tool Schema**: Each tool defines its input and output schemas (usually with JSON Schema or Zod). - **MCP Protocol**: JSON-RPC 2.0 based, with methods like `tools/list` and `tools/call`. - **Mocking**: For testing, all external dependencies (e.g., Float API) must be mockable. - **Error Handling**: All errors must be structured and informative. --- ## Server Structure - `src/tools/`: Tool definitions and handlers - `src/services/`: Service integrations (e.g., Float API) - `src/config/`: Configuration and environment management - `tests/`: Comprehensive test suite (integration, unit, mocks) --- ## Key Endpoints - `/tools/list`: List all available tools and their schemas - `/tools/call`: Invoke a tool with arguments --- ## Best Practices - All tools must validate input and output. - Use Zod for schema validation. - Ensure all tools are discoverable via `/tools/list`. - Provide clear error messages and codes.