cognee vs hindsight
The two are close in size: 31.0k stars for cognee, 30.5k for hindsight. Over the days we have tracked them cognee moved +21.6% and hindsight +72.7%, so hindsight is growing faster right now.
They split the axes: cognee leads on context depth, hindsight on setup ease.
Stars and commit dates come from our own daily tracking. The six axes are read off each project's documentation by our review pipeline, so they describe what a project says about itself, not what we measured in its code.
Where they stand today
Runs the full memory layer (graph, vectors, sessions) on a single self-hosted Postgres-backed stack, combining vector search and evolving knowledge graphs.
- Stars
- 31.0k
- Tracked growth
- +21.6%
- Maturity
- ●●●●●
- Last commit
- 9h ago
- Language
- Python
- License
- Apache-2.0
- Cost to run
- Your LLM API key, or use Cognee Cloud (managed)
Hindsight provides a learning-focused agent memory with biomimetic data structures and a reflect operation to build mental models, going beyond simple RAG-style recall.
- Stars
- 30.5k
- Tracked growth
- +72.7%
- Maturity
- ●●●●●
- Last commit
- 14h ago
- Language
- Python
- License
- MIT
- Cost to run
- Your LLM API key (or self-hosted models); Hindsight itself can be self-hosted
Six axes, head to head
Each axis runs 0 to 5. The label under a score is what that project's own docs claim, not a category average.
| Axis | cognee | hindsight |
|---|---|---|
Context depth How much of your codebase it sees before it answers: the open diff, the diff plus related files, or the whole repository. | ●●●●● Whole-graph memory | ●●●●● Related context |
Noise control How it keeps output volume down — severity thresholds, deduplication, incremental runs over new commits only. | ●●●●● Feedback & routing | ●●●●● Rerank & filtering |
Customization How far it bends to your team: custom rules, prompts, style guides, per-path config. | ●●●●● Config + plugins | ●●●●● Config API & metadata |
Privacy Whether your code stays on your own infrastructure: fully local, self-hostable, or cloud API only. | ●●●●● Self-hostable | ●●●●● Self-hostable |
Model freedom Whether you can point it at any provider, or it is wired to one. | ●●●●● Bring-your-key | ●●●●● Bring-your-own / local |
Setup ease What it takes to get a first useful run out of it. | ●●●●● API key required | ●●●●● One-command start |
Which one to pick
Pick cognee if…
Self-hostable — run a unified, persistent agent memory on a single Postgres instance with minimal extra services and strong integration options.
- Context depth: Whole-graph memory (5/5 against 3/5)
Pick hindsight if…
Learning-focused — choose Hindsight when you need agents that form and reflect on long-term memories (mental models) with easy self-hosted deployment and multi-provider model support.
- Setup ease: One-command start (5/5 against 3/5)
What people want from each one
topoteretes/cognee
Hacker News: Show HN: Cognee – Open-Source AI Memory Layer That Remembers Context drew 9 points and 2 comments.
vectorize-io/hindsight
- Feature Request: Team, agent, and memory-level access control 👍 5
- reranker(typesafe): every recall fails over on banks with realistic-size memories — the request ships full candidate text against Jev's ~32k budget, and nothing bounds it 👍 2
- [Feature]: I'd like to benefit from disk savings with Embedding Quantization 👍 1
Questions people ask
Is cognee better than hindsight?
They split the axes: cognee leads on context depth, hindsight on setup ease. cognee is worth picking when self-hostable — run a unified, persistent agent memory on a single Postgres instance with minimal extra services and strong integration options.
Which of cognee and hindsight keeps my code private?
cognee: Self-hostable (4/5). hindsight: Self-hostable (4/5).
What does each one cost to run?
cognee: Your LLM API key, or use Cognee Cloud (managed). hindsight: Your LLM API key (or self-hosted models); Hindsight itself can be self-hosted.
Full profiles: topoteretes/cognee and vectorize-io/hindsight. Everything else in Memory & context.