cognee
https://github.com/topoteretes/cognee
Python
Memory for AI Agents in 5 lines of code
Triage Issues!
When you volunteer to triage issues, you'll receive an email each day with a link to an open issue that needs help in this project. You'll also receive instructions on how to triage issues.
Triage Docs!
Receive a documented method or class from your favorite GitHub repos in your inbox every day. If you're really pro, receive undocumented methods or classes and supercharge your commit history.
Python not yet supported0 Subscribers
Add a CodeTriage badge to cognee
Help out
- Issues
- Fix: PGVectorAdapter never creates an ANN index
- LanceError: Spill has sent an error — also reproduces out-of-process and at 384/512-dim (not just uvicorn + 1024-dim)
- fix: Return single task result from run_tasks_parallel
- [Bug]: run_tasks_parallel returns an empty list when given a single task
- LanceDB merge_insert 'Spill has sent an error' with 1024-dim custom embeddings — only inside uvicorn server; same pipeline succeeds out-of-process
- Graph visualization hangs forever at "Laying out graph... 0%" when source text has unbalanced HTML comments (<!-- without matching -->)
- [Bug]: recall ratifies affirmative/keyword queries as stored facts — and can harden them (e.g. 'added to watch list' becomes 'merged')
- Skills page silently truncates at 200 skills — frontend sends no limit/offset while `GET /v1/skills` defaults to `limit=200`
- `POST /v1/skills` returns generic 409 "Failed to ingest skill" for every exception, masking unrelated ingest bugs
- `/api/v1/visualize` takes ~197 s for an 828-node graph; same graph via `datasets/{id}/graph` returns in 0.23 s
- Docs
- Python not yet supported