Your agents share the work.
Now they share the memory.
Apache-2.0 · single-binary self-host · works with MCP, LangGraph, CrewAI, Claude Agent SDK
npm i @loregpt/sdk Lore is the open-source coordination memory layer for multi-agent AI systems — read-your-writes consistency, built-in access control, and a token bill that goes down.
The coordination problem
Why multi-agent systems fail
They mostly don’t fail because agents can’t reason. They fail because agents work over inconsistent copies of shared state.
The invisible handoff
Your researcher found it. Your writer never saw it. Every agent starts from zero — and your token bill proves it.
Stale truth
Two agents updated the same fact. Which one is your system serving now? If you can’t answer, neither can your agents.
Debugging in the dark
“Why did the agent do that?” shouldn’t take an afternoon. Without provenance, every incident is archaeology.
Everything between your agents and one shared reality
The pieces you’d wire yourself — and the layer Lore handles.
Compatible with leading AI clients
Lore ships as native SDKs and a single-binary MCP server — so it drops into the frameworks and editors your agents already run in.
Frameworks & SDKs
- Claude Agent SDK
- LangGraph
- CrewAI
- OpenAI Agents SDK
- MCP
Editors & agents · via MCP
- Cursor
- Windsurf
- Cline
- Qwen Code
Plus support for any MCP-compatible AI tool — framework-neutral by design.