# Engrim

21 Hacker News commenters recommended Engrim and 0 criticized it, November 2025 to November 2026. It was recommended in 4 of 13 months. It meets the Hacker Approved badge rule.

**Verdict:** Open-source local memory engine for AI agents, built on SQLite, with local embeddings and no API key or external service needed. Critics note it is still limited by effective context size.

## Why people recommend it

- Local-first, no API key needed (6 comments): "without requiring a vector database, an OpenAI key, or any external service" ([source](https://news.ycombinator.com/item?id=47307500)) "Local SQLite is practically zero-latency, keeps your IP entirely private" ([source](https://news.ycombinator.com/item?id=49597713)) "Local-first + SQLite is the right call for offline-first agent memory" ([source](https://news.ycombinator.com/item?id=49595488))
- Brain-inspired memory with decay and consolidation (6 comments): "Instead of vector DB + semantic search, it uses models from cognitive science (ACT-R activation decay, Hebbian learning, forgetting curves)" ([source](https://news.ycombinator.com/item?id=47351473)) "modeling things like salience, decay, and consolidation is much closer to how useful memory actually works" ([source](https://news.ycombinator.com/item?id=47037794)) "The result is agents that remember context across sessions, learn from patterns, and forget gracefully." ([source](https://news.ycombinator.com/item?id=47226181))
- Remembers context across sessions and agents (4 comments): "The result is agents that remember context across sessions, learn from patterns, and forget gracefully." ([source](https://news.ycombinator.com/item?id=47226181)) "agents that remember context across sessions, learn from patterns, and forget gracefully" ([source](https://news.ycombinator.com/item?id=47228096)) "an MCP server that gives all your agents a shared, persistent knowledge base" ([source](https://news.ycombinator.com/item?id=47602486))
- Records decisions and their provenance (4 comments): "tracking which agent (or human) made which decision ended up being one of the most useful parts of the whole system for me" ([source](https://news.ycombinator.com/item?id=49597956)) "separating settled decision records from WIP state tells the new agent what was finished vs what died mid-flight" ([source](https://news.ycombinator.com/item?id=49599677)) "I especially like the provenance idea." ([source](https://news.ycombinator.com/item?id=49597893))

## What people warn about

- Still limited by effective context size (1 comments): "It's an improvement, but obviously you're still limited with effective context size" ([source](https://news.ycombinator.com/item?id=49606181))

## Alternatives hackers recommend

- [MemPalace](https://hackerapproved.com/p/mempalace.md): 9 recommend
- [Grov](https://hackerapproved.com/p/grov.md): 8 recommend
- [claude-mem](https://hackerapproved.com/p/claude-mem.md): 8 recommend
- [Letta](https://hackerapproved.com/p/letta.md): 8 recommend
- [Hippo AI agent memory](https://hackerapproved.com/p/hippo.md): 8 recommend

## Cite as

Hacker Approved, "Engrim", https://hackerapproved.com/p/engrim/ (counts from 1,795,640 Hacker News comments, November 2025 to November 2026).