At a glance
1. Add knowledge
Bring in documents, notes, repositories, meetings, and connected sources.
2. Build context
Index content, extract useful links and facts, and keep related pages connected.
3. Help agents act
Let compatible agents search the brain and follow the guidance you approve.
The knowledge loop
1
Collect source material
Start with the information your team already uses: Markdown files, repository documentation, meeting notes, and imported records. A brain can contain more than one source, so different workspaces can stay separate while still being searchable when appropriate.
2
Make it searchable and connected
ModusBrain stores pages, chunks content for search, and can extract links, entities, timelines, and facts. This gives a question more than keyword matches: it gives it the surrounding context.
3
Retrieve the right context
When a person or agent asks a question, ModusBrain combines keyword, vector, and graph-aware retrieval. Results stay scoped to the correct brain and source.
4
Use approved operational guidance
For repeatable or sensitive work, compile an operational skill from the knowledge in the brain. Skills can be reviewed, approved, confidence-gated, and audited.
5
Keep the brain current
Sync new source changes, enrich pages, and run maintenance when needed. The result is a brain that improves as your knowledge changes.
Choose your path
Use it locally
Set up a local PGLite brain with the CLI.
Connect an agent
Serve your brain through MCP for coding agents and other clients.
Run it for a team
Use Postgres, scoped sources, and access controls for shared company knowledge.
Explore the architecture
Learn about brains, sources, storage engines, and deployment shapes.