Garry Tan open-sources gbrain-evals, a public test suite for agent memory systems
garrytan · x · 2026-10-11
Garry Tan open-sourced gbrain-evals (434 stars, MIT): gbrain is a memory system for AI agents, and gbrain-evals is its public test suite with data, comparison systems, scoring code, and all published results — reusable against other memory systems. He also reports agents auto-improving retrieval/latency evals via Capy with zero human input for 16 hours.
Related event: YC's Garry Tan Shares His Agentic Coding Workflow(4 posts)→
More from coding & agent
- GhidraMCP updated for latest Ghidra with headless support: load plugin and start reversing — lauriewired · 2026-10-11
- 13-year gamedev builds Claude Code-driven 3D asset pipeline, barely writing code by hand — Embarrassed_Guide_80 · 2026-10-11
- Harvard Med workshop teaches building AI co-scientists with ToolUniverse's 2,700+ tools — marinkazitnik · 2026-10-11
- Delvetown agents keep producing impressive artifacts as AI town experiment evolves — lfschiavo · 2026-10-11
- Six public DNS signals agents can use to verify emails and domains in one second — S_B_B_B_K · 2026-10-11
- Nous Research launches stealth coding and agentic reasoning model, free for limited time — Teknium · 2026-10-11