topoteretes / cognee
Code health
This codebase scores 6.6 out of 10 for code health, which we rate fair. It also scores maintainability 8.1 and static performance 9.8 out of 10. The three are scored separately and never blended into one number. Risk is concentrated, as it usually is: 298 of 2,262 files are git hotspots, and they average 4.6 — which is where the fixes pay off most.
Do next
3 things worth doing this quarter, 2 of them now.
Worth planning
Delete 126 unused symbols and files (2,856 lines)
Nothing in the graph reaches them, across 99 files; every reader and agent pays to skip them.
- symbols
- 126
- lines
- 2,856
- safe to delete
- Yes(inferred)
Medium effort
Improve what Repowise can see
Add a test coverage report
Without one, Repowise cannot tell tested code from untested code, so every test-related action says “unknown”.
- line coverage
- Unknown, no report
Small effortReview the 59 proposed decisions
None is accepted yet, so nothing can drift from one and agents get no enforced guidance from them.
- proposed
- 59
- accepted
- 0
Medium effort
- Lines of code
- 254.7K
Recent activity
All commitsCommits
- af39184edocs(evals): archive legacy HotpotQA suite3mo ago
- 524ff4b2chore(release): bump version to 1.4.03mo ago
- 7c7144d1fix(improve): pass user to add/cognify in agent trace feedback persistence (COG-5893)3mo ago
- 8c6cebfcrefactor(ingestion)!: remove the in-core Gmail connector (moved to cognee-community)3mo ago
- 8499ece0feat(ingestion): respect connector write_disposition for document sources + always graph-clean orphans3mo ago
- 33ced74cfeat(ingestion): connector-agnostic DLT document-mode + foreground orphan cleanup3mo ago
Decisions
- Ensure stored get_graph_engine() references remain live across operationsproposed · 2mo ago
- Avoid LiteLLM cold import just to set verbose=Falseproposed · 2mo ago
- Suppress external library loggers by name before importproposed · 2mo ago
- Suppress LiteLLM logging via env vars even if litellm not imported yetproposed · 2mo ago
- Avoid importing package init to get versionproposed · 2mo ago
- Persist agent trace feedbacks into knowledge graph and extend agent-memory/session-trace workflowproposed · 2mo ago
Everything we found
6,599 open
Every detector's open count by area, each with its worst example. An inventory, not a to-do list: the actions above are the part worth doing.
- worst item is critical severity. Code health4,299Untested hotspot · cognee/infrastructure/databases/graph/ladybug/adapter.py80 findings · 10.0 deducted
- worst item is medium severity. Refactoring383imgIcon+1.5 health, S effort · cognee-frontend/src/app/(app)/integrations/page.tsx
- worst item is medium severity. Decisions357cognee/infrastructure/session/session_manager.py59 awaiting review · 298 hotspots ungoverned
- worst item is medium severity. Ownership1,394cognee/api/v1/remember/remember.py85% single-owner concentration, 32 commits in 90d
- worst item is low severity. Dead code166cognee/modules/retrieval/context_providers/DummyContextProvider.pyUnreachable symbol (20 lines)
Composition
Open the graph- python88%
- typescript11%
- javascript<1%
- shell<1%
- dockerfile<1%
Explore this codebase
- Docs0 pages across 15 modules
- ChatAsk this codebase a question and get an answer with its sources.
- Files2,262 files with per-file docs, health and history
- ArchitectureDependency graph, layers, and 16 entry points
- Code healthPer-file scores, 5,096 open findings, coverage and refactoring targets
- Knowledge graphEntities, communities, and the paths between them
- Change couplingFiles that keep changing together, mined from commit history
- CommitsChange-risk ranked history with agent provenance
- ContributorsBus factor, per-file maintainers, and the human/agent split
- StatsSize class, origin, lifetime churn, rhythm and records
- CostsWhat indexing this snapshot cost, by model and by run.