What it does
Beever Atlas pulls conversations from Slack, Discord, Microsoft Teams and Mattermost, extracts atomic facts, removes duplicates and groups them into topic pages with citations. A graph store links the people, decisions and projects mentioned across channels. A six-stage pipeline builds this memory, and each channel gets its own wiki that is rebuilt as new messages arrive.
Teams reach for it when they want questions about past discussions answered in natural language, with each answer citing its source messages. Answers can be asked through the dashboard or through the MCP server, which connects Claude Code and Cursor to the knowledge base. The MCP server exposes 28 tools with per-agent authentication. Relational questions, such as who worked on a topic with whom, are routed to the graph store.
The stack runs as Docker Compose services: backend, bot and frontend, backed by Weaviate, Neo4j, MongoDB and Redis. A seeded demo with a public Wikipedia corpus can be started with make demo, and seeding needs no keys. Asking questions through the API requires a free-tier GOOGLE_API_KEY, because the QA agent calls Gemini.
Limitations and requirements
Asking questions through the API requires a free-tier GOOGLE_API_KEY, because the QA agent calls Gemini. The full installation requires API keys, and the stack needs Docker Compose with four datastores (Weaviate, Neo4j, MongoDB and Redis). Chat platforms must be connected to the bot before their history can be synced.
Safety
Our team has not reviewed this resource. Only automated rules checked it; resources with high-risk signals are not published. These are signals, not a guarantee that installing it is safe. Try it in a sandbox project first.
- Runs code or changes an assistant's behaviour
Sources
- Official MCP Registry first seen 13 Nov 2026 · last seen 13 Nov 2026 · its page on the source
io.github.Beever-AI/beever-atlas
Open-source LLM knowledge base: team chat into a typed knowledge graph and auto-generated wiki.
v0.3.0As shown on registry.modelcontextprotocol.io when we last read it