How to ask questions answered by 📝RAG over your 📝MythOS knowledge library from any 📝MCP-connected AI client.

When your library grows beyond what you can hold in your head, chat_with_library lets your AI search semantically across your memos and answer questions grounded in your own knowledge — not its training data. See: 📝MCP vs RAG.

What You'll Need

• A MythOS library connected to an AI client via MCP (see setup tutorials for 📝Claude.ai, 📝Claude Code, or 📝Cursor / Claude Desktop)

Steps

1. Ask — a question in natural language that relates to your library content. Example: "What does my library say about distributed systems?" Your AI calls chat_with_library behind the scenes
1. Review — the response. The AI retrieves semantically relevant memos via vector search and grounds its answer in your content
1. Follow up — in the same conversation to continue the thread. The AI maintains context across turns within the same thread
1. Verify — by checking the memos cited. If something's missing from the answer, create a memo for it — your library is only as good as what you put in

RAG-powered chat respects your library's permission model. Private memos are only retrievable by you. Memos with audience tags are filtered based on the requester's relationship to the library owner.

What's Next?

• 📝MCP vs RAG — understand the distinction between connectivity and retrieval
• 📝MythOS Content Topology — structure your library for better retrieval
• 📝MythOS MCP — full MCP reference including all tools

Contexts

• 🏷️#mythos
• 🏷️#mythos-feature