Read 2026-10-23
topoteretes/cognee: technical due diligence, read from its repository
What Borehole read in topoteretes/cognee's code and history at 37f788b: its licence, security advisories, releases, review practice, tests, dependencies and operations. Every finding cites the file and line it rests on. The full report also covers the team and asks the questions the code cannot answer.
| Dimension | Rating |
|---|---|
| Architecture & Codebase — including AI provenance | adequate |
| Engineering Process & SDLC Maturity | strong |
| Product & Engineering Maturity | strong |
| Security & Reliability Posture | adequate |
| Observability & Operations | strong |
| Fit with the Acquirer | not assessable |
What the repository says
CONTRIBUTING.md:153: The project does not take pull requests- Releases are cut from another branch than the one surveyed
.github/pull_request_template.md:6: states a policy on AI-generated contributionsSECURITY.md:158: names OpenSSF
Architecture & Codebase — including AI provenance
adequatecognee/infrastructure/databases/graph/ladybug/adapter.py is 3,972 lines, 44× this codebase's median file
adequateThe median source file here is 89 lines. 26 files exceed 1,000 lines, the largest being 3,972. Size is not a defect on its own — some problems genuinely live in one place — but these are where merge conflicts, review fatigue and single-owner knowledge concentrate, and they are the first thing that slows an inheriting engineer down.
cognee/infrastructure/databases/graph/ladybug/adapter.py:1 — 3,972 lines (44× median) |
cognee/modules/retrieval/code_retriever.py:1 — 3,262 lines (36× median) |
cognee/infrastructure/databases/graph/neo4j_driver/adapter.py:1 — 2,550 lines (28× median) |
cognee/modules/visualization/views/story_view.js:1 — 2,354 lines (26× median) |
cognee/api/v1/remember/remember.py:1 — 2,211 lines (24× median) |
4 of 6 notebooks are committed with their outputs
adequate67% of the notebooks carry cell outputs in git, and nothing strips them before commit. Outputs bloat the history, make changes unreadable in review, and carry whatever the cells printed: data, and sometimes keys.
notebooks/cognee_demo.ipynb — notebook with outputs |
notebooks/cognee_multimedia_demo.ipynb — notebook with outputs |
notebooks/cognee_simple_demo.ipynb — notebook with outputs |
Engineering Process & SDLC Maturity
strongTests are retried until they pass
adequateThe suite is configured to retry failing tests (cognee-frontend/src/services/http/__tests__/client.test.ts). A retry turns a flaky test green without saying why it failed, so a green build means the tests passed eventually. Ask how many retries a typical run needs, and which tests need them.
cognee-frontend/src/services/http/__tests__/client.test.ts:299 — test retry |
Product & Engineering Maturity
strongReleases are cut from another branch than the one surveyed
adequateThe newest pre-release, v1.6.6rc1, is not on the branch this survey read; the newest release on it is v1.6.5. The shipped product may differ from what this report describes. Survey the release branch too.
| tag v1.6.6rc1 |
Security & Reliability Posture
adequate1 Dockerfile runs the application as root
adequateNo USER instruction switches away from root, so the process runs with full privileges inside the container. It is a one-line fix and an ordinary expectation, which is why its absence is worth asking about: it usually says more about whether anyone has reviewed the deployment than about this container specifically.
cognee/eval_framework/Dockerfile — runs as root |
11 SQL statements built from strings
adequate11 places in product code build SQL by formatting or joining strings rather than passing parameters, and in 8 of them a value, not a table or column name, goes into the string. Where any of those values comes from a request, it is an injection point. Each site needs a look; the fix, parameters, is small.
cognee/infrastructure/databases/graph/turso/adapter.py:291 — SQL built from a string |
cognee/infrastructure/databases/graph/turso/adapter.py:309 — SQL built from a string |
cognee/infrastructure/databases/graph/turso/adapter.py:597 — SQL built from a string |
4 shell commands built from strings
adequate4 places run a command through a shell (`shell=True`, `os.system`, `exec` with a template) rather than passing arguments. Where any part comes from outside, it is command injection. Ask where each one's input comes from.
cognee/api/v1/ui/node_setup.py:41 — command run through a shell |
cognee/api/v1/ui/node_setup.py:291 — command run through a shell |
cognee/api/v1/ui/npm_utils.py:19 — command run through a shell |
38 locked dependencies have published vulnerabilities, 58 high or critical in what ships
adequate38 of 944 locked dependencies match 140 published advisories in OSV, 65 of them rated high or critical. A lockfile says which version is installed, not whether the flawed code is reachable from this product, so each needs a look. Most are fixed by an upgrade, and the advisory names the version.
cognee-mcp/uv.lock — anyio 4.14.1: GHSA-82r6-8w77-94w6 (critical), fixed in 4.14.2 |
cognee-mcp/uv.lock — certifi 2026.6.17: GHSA-6q48-6c7w-9gw2 (critical), fixed in 2026.10.19 |
uv.lock — certifi 2026.7.22: GHSA-6q48-6c7w-9gw2 (critical), fixed in 2026.10.19 |
5 places insert HTML from a variable without sanitising it
adequate5 places set HTML from a value that is neither a literal nor passed through a sanitiser. Where the value can come from a user, that is stored cross-site scripting, and it is the first thing a buyer's security review looks for in a frontend.
cognee-frontend/src/modules/config/RuntimeConfigScript.tsx:18 — HTML from a variable |
cognee/modules/visualization/views/semantic_map.js:219 — HTML from a variable |
cognee/modules/visualization/views/story_view.js:1196 — HTML from a variable |
3 advisories against this project in the last 24 months, 2 high or critical
adequateThe newest was published on 2026-10-04. A project that publishes advisories is telling its users about its fixes, and the count and pace say how much security work it carries. None of them affects the surveyed version.
| 2026-10-04: CVE-2026-105141 (moderate): A security flaw has been discovered in topoteretes cognee up to 1.5.4. The affected element is the function get_user_id_by_email of the file cognee/modules/user |
| 2026-07-07: GHSA-49f7-whx5-4256 (critical), fixed in 1.5.0: Cognee allows non-superusers to overwrite global LLM configuration |
| 2026-05-12: GHSA-8pr4-p4c8-8gr9 (critical), fixed in 1.2.0: Cognee notebook cell execution endpoint passes request bodies to exec(), allowing remote code execution |
Observability & Operations
strong1 of 2 deployments run a single replica
adequate1 deployments set `replicas: 1`, and nothing (an autoscaler, a disruption budget) adds more. Every deploy and every node drain is downtime, and the availability the buyer inherits has a ceiling.
deployment/helm/templates/postgres_deployment.yaml — replicas: 1 |
Fit with the Acquirer
not assessableNot rated: structurally not assessable from a repository.
Read more, or read your own
The full report on topoteretes/cognee: every dimension, the questions for the room, and the audit log.
Survey a repository: free on public repositories. Private code is read on your own machine by a collector whose source you can read first.