GBrain
garrytan/gbrain
Agent memory that stores sourced facts and shares them across agents over MCP, built by Garry Tan
GBrain is an open-source memory system that stores explicit facts with their sources and makes the same memory available across several agents. Its author describes it as the production brain behind his own OpenClaw and Hermes deployments.
What GBrain does
GBrain stores explicit facts with their sources, supports corrections and withdrawal, and exposes the same memory to different agents. Setup starts with keyless memory and keyword retrieval; semantic search, synthesis and background enrichment can be added later. New local brains include memory skills in their content root, and agents can join an authorized catalog where only approved editors can publish.
The README offers three setups: add GBrain to an existing agent through guides for Grok Bot, Muse and Codex or Claude Code; run the brain on your own computer and publish it over MCP on a Tailscale tailnet with gbrain mcp expose; or connect an existing hosted brain. Optional features include a synthesis layer that returns cited answers with a note on what the brain does not know yet, and a typed knowledge graph. It can serve as a company brain too, but the README warns that local files and shared database credentials sit outside the access controls for remote clients.
Key features
- Stores explicit facts with sources, with corrections and withdrawal
- Keyless memory with keyword retrieval, plus optional semantic search and synthesis
- Publishes the brain over MCP with gbrain mcp expose, including Tailscale access
- Optional synthesis layer that returns cited answers and gap analysis
- Optional typed knowledge graph
- Memory skills kept beside knowledge, shared through an authorized catalog
When to use it
- Giving an existing agent a memory you control and can correct
- Sharing one memory between Hermes, Codex, Claude Code and other agents over MCP
- Running a company brain with source and operation grants for remote clients
Who it is for: People who run one or more agents and want an explicit, portable memory with provenance.
How it fits with Hermes Agent
The repository description calls it an OpenClaw/Hermes Agent brain, and the author says it is the production brain behind his OpenClaw and Hermes deployments. The setup guides also cover Grok Bot, Muse, Codex and Claude Code, and other harnesses connect over MCP.
FAQ
What is GBrain?
GBrain is a memory system for AI agents that stores explicit facts with their sources and shares the same memory across agents. It supports corrections, withdrawal and optional synthesis of cited answers.
Does GBrain work with Hermes Agent?
Yes. The repository describes itself as an OpenClaw/Hermes Agent brain, and the author says it runs behind his own OpenClaw and Hermes deployments. Other harnesses connect over MCP.
Is GBrain free and open source?
Yes, the repository is released under the MIT license.
Similar memory for Hermes Agent
All memoryMemory and context engine for AI agents with an MCP server, open-source plugins and a Hermes provider
TencentCloud TencentDB Agent MemoryTeam-level memory hub that shares chat memory, skills, wiki and code graph across agents through a proxy
screenpipe screenpipeRecords screen and audio locally and exposes searchable computer history to agents via CLI skills and MCP
MemoriLabs MemoriLLM-agnostic memory layer with Python and TypeScript SDKs that persists and recalls agent conversations
NevaMind-AI memUShared LLM wiki memory across sessions, agents and devices that also distills reusable skills
MemTensor MemOSMemory operating system for LLMs and agents with official local plugins for Hermes Agent and OpenClaw
Related guides: SOUL.md for Hermes Agent: what it is and how to write one · Run multiple Hermes agents with profiles