FREQUENT QUESTIONS
Before you start using MESA.
Direct answers about what you install, what it does and what it needs.
What is MESA?
A tool that stores and searches documents for the AI application you build. It returns relevant text with the document it came from.
Who is it for?
Developers building an assistant, a research tool or an agent that needs to use saved information. MESA is not a ready-made chat application.
Does MESA write the answer?
No. It finds the text your application needs. Your application and its language model use that text to write the answer.
Do I need a model account?
Not for basic word search. Add a model provider for meaning search; also set MESA_EXTRACTION_ENABLED=true for model-based facts and relationships. Legal rule items work without a model.
How do I try it?
Install the Python package, run mesa start, then add a text file and search it. The installation guide lists the commands and supported systems.
Does MESA have a CLI?
Yes. The mesa command comes with the mesa-memory package. Start the server with mesa start; add, search and delete documents with mesa add, mesa search and mesa delete. Use mesa status to check the server.
Can I connect an MCP client?
Yes. Run mesa start, then mesa mcp-config. Keep the generated connection settings private because they contain your local API key.
What if the document changes?
After a new revision is processed, it replaces the old revision in current search results. Document update and deletion operations are described in the API guide.
Is MESA production-ready?
Not yet. Version 0.9.3 is published, but the required production qualification is incomplete. Try it locally and read the current limits before integrating it.
