Also, I want to know backend stack using OpenHands. Or User need to setup his own stack for agents to: • understand the entire repo via RAG / vector DB, • navigate the structure using an architecture map (Repomap or equivalent), • interact with all tools via MCP (Model Context Protocol), • automate tasks like tests, builds, PRs, and refactorings. I’d like to know if in OpenHands: 1. There’s already an integrated backend or MCP connectors for: ◦ vector DB / RAG (embedding extraction, semantic search), ◦ code indexing / semantic search (like Cody or Sourcegraph), ◦ generating or importing repo maps for architecture. 2. There are ready-to-use MCP tools to expose these services to OpenHands agents. 3. Any examples of complete backend architecture or “blueprints” for OpenHands exist to handle large repos and multiple agents in pipeline or parallel. If these components don’t exist yet, does the OpenHands team have practical recommendations for implementing them easily, ideally: • creating MCP tools for vector DB, map, and search, • orchestrating multiple agents on the same repo, • managing full context for projects >300k lines. Any documentation, architecture diagrams, or example packages would be super helpful. Thanks in advance 🙏 #C0ACTL5GAQ3

OpenHands supports MCP, so please feel free to add the capabilities and try them!