Reddit MCP Server

# Brainloop MCP Server Source Code This directory contains the source code for the Brainloop Model Context Protocol (MCP) server. The server enables AI assistants to interact with Brainloop through a standardized protocol. ## Directory Structure ### Entry Points - **`index.ts`** - Main executable entry point for running the server directly - **`server.ts`** - HTTP server implementation that handles OAuth and hosts MCP endpoints ### Core Directories #### `/constants` Static definitions for tools, sampling operations, and their schemas. This is where all tool definitions and prompts are centralized. #### `/handlers` Request handlers that implement the business logic for: - Tool execution (create, view, manage brainloops) - Sampling operations (AI-assisted content generation) - Notifications and progress tracking - Resource management #### `/server` HTTP server infrastructure including: - OAuth2 authentication flow (Google OAuth) - MCP protocol endpoints - Session management - Middleware for security and validation #### `/services` External service integrations: - Brainloop API client with OAuth support #### `/types` TypeScript type definitions for: - Brainloop API data structures - MCP protocol extensions - Internal application types #### `/utils` Utility functions for: - Validation - Logging - Error handling ## Architecture Overview The server follows a layered architecture: 1. **Entry Layer** - Main entry point for server execution 2. **Server Layer** - HTTP server with OAuth and MCP protocol support 3. **Handler Layer** - Business logic for processing MCP requests 4. **Service Layer** - Brainloop API integration 5. **Utility Layer** - Cross-cutting concerns ## Key Concepts ### Authentication Flow 1. User initiates OAuth flow through `/oauth/authorize` 2. Google redirects back with authorization code 3. Server exchanges code for access token 4. Token is stored per session for subsequent API calls ### MCP Protocol Implementation - Tools allow AI to create, view, and manage brainloops (courses) - Sampling enables AI-assisted content generation - Notifications provide real-time feedback - Sessions maintain authentication context ### Multi-Session Support The server supports multiple concurrent users, each with their own: - Authentication credentials - MCP server instance - Session state ## Development To understand how the server works: 1. Start with `index.ts` to see the main entry point 2. Follow the flow through `server.ts` for HTTP setup 3. Look at `/server/mcp.ts` for MCP protocol handling 4. Examine `/handlers` for business logic 5. Check `/services` for Brainloop API integration Each subdirectory contains its own README with more detailed information about its specific functionality.