Engram: MIT-licensed local-first memory for coding agents with SQLite FTS5, BM25 and no network at recall
utsapoddar · reddit · 2026-10-03
The author released Engram, a free MIT-licensed local-first memory layer for coding agents:
- Memories are plain Markdown files on disk
- Search uses BM25 over a SQLite FTS5 index rebuilt from the Markdown, so the index is disposable
- If a local embedding model exists, cosine results are fused with lexical results via reciprocal rank fusion; recall never touches the network or downloads models
- The test suite enforces recall@5 ≥ 90% across 20 seeded queries, acting as a regression gate rather than a benchmark
Repo: github.com/utsapoddar/engram
More from coding & agent
- Engineer builds production-grade PCB in 3 weeks with AI, works on first revision — cneuralnetwork · 2026-10-03
- Jev open-sourced: PowerShell module turns natural language into typed AI decisions — dfinke · 2026-10-03
- Video explains Jev agent examples: games, triage, and browser use — RelevantEmergency707 · 2026-10-03
- Making Claude music videos: upload your own audio instead of browser-driving Suno — technollama · 2026-10-03
- 'AI Babysitter' Is the New Job: Keeping Agent Queues at 100% Utilization — petesena · 2026-10-03
- Antigravity CLI 1.2.15 runs the full coding agent on Android via Termux — kalyan_kpl · 2026-10-03