GBrain is an open-source agent memory layer from Y Combinator CEO Garry Tan that gives the agents you already use a memory you control.
It stores explicit facts with sources, supports corrections and withdrawal, and shares the same memory across Claude Code, Codex, OpenClaw, Hermes and other harnesses.
- 01
Explicit fact storage with provenance, corrections and withdrawal
- 02
Keyword retrieval out of the box, with optional semantic search across many embedding providers
- 03
Synthesis layer that returns cited answers and notes knowledge gaps
- 04
Typed knowledge graph with automatic reference extraction
- 05
MCP server exposure over Tailscale (gbrain mcp expose) for local and cloud agents
- 06
Skill packs and a native router for coding-agent installs such as Claude Code and Codex
- 07
Shared company-brain mode with source and operation grants
Open Source
GBrain is MIT-licensed and free to install with Bun. It starts with keyless memory and keyword retrieval; optional semantic search, synthesis and background enrichment use embedding and LLM providers you configure (including local options such as llama.cpp and LM Studio), which may incur their own costs.
Best for
Developers and power users running several AI agents who want one durable, cited memory shared across harnesses and machines.
Not ideal for
Users wanting a simple zero-config plugin: setup has many options and trust boundaries to understand, and the npm package named gbrain is unrelated, so installation must follow the documented Bun path.
GBrain is a powerful, MIT-licensed memory layer proven in its author's own large OpenClaw and Hermes deployments, and increasingly usable from coding agents. It rewards users willing to invest in setup; those wanting lightweight session memory for a single coding agent may prefer simpler plugins.
Very actively developed with near-daily releases (v0.60.x in October 2026) and trending on GitHub, with guides for Codex, Claude Code, Grok and Muse harnesses and hosted MCP access.
As of
Is GBrain free?+
Yes - GBrain is Open Source. GBrain is MIT-licensed and free to install with Bun. It starts with keyless memory and keyword retrieval; optional semantic search, synthesis and background enrichment use embedding and LLM providers you configure (including local options such as llama.cpp and LM Studio), which may incur their own costs.
Is GBrain open source?+
Yes - GBrain is open source. GBrain is MIT-licensed and free to install with Bun. It starts with keyless memory and keyword retrieval; optional semantic search, synthesis and background enrichment use embedding and LLM providers you configure (including local options such as llama.cpp and LM Studio), which may incur their own costs.
Who is GBrain best for?+
Developers and power users running several AI agents who want one durable, cited memory shared across harnesses and machines.
Who is GBrain not ideal for?+
Users wanting a simple zero-config plugin: setup has many options and trust boundaries to understand, and the npm package named gbrain is unrelated, so installation must follow the documented Bun path.
What are the best GBrain alternatives?+
The closest GBrain alternatives on ai.dosa.dev are Pieces for Developers, agenttrace, Vibe Kanban - all listed under Developer Productivity & Workflow.
Who makes GBrain?+
GBrain is developed by Garry Tan. It is listed in the Developer Productivity & Workflow category on ai.dosa.dev.
Favorite this tool to revisit it later, or Zap it to contribute to the public vote count.