DocsLearn
Learn
Shared across agents
Every agent you connect reads and writes one file, ~/.memorylayer/memories.db. A decision made in Cursor today is there when Claude Code asks tomorrow.
Two ways to connect
Two ways to connect
SSE one shared server
Claude CodeCursor…
memorylayer startlocalhost:7400/sse
memories.db~/.memorylayer
stdio no server to run
Your agent
memorylayer --stdiokey via MEMORY_API_KEY
memories.dbsame file
stdio — what setup writes
The agent starts MemoryLayer itself and passes your key in the environment.
{"mcpServers": {"memorylayer": {"command": "memorylayer","args": ["--stdio"],"env": { "MEMORY_API_KEY": "sk-ml-your-key-here" }}}}SSE — one server for every agent
Start the server and leave it running (memorylayer --key on its own only saves the key), then point each agent at it:
memorylayer start --key sk-ml-your-key-herecurl http://localhost:7400/health # → {"status":"ok","transport":"sse",...}{"mcpServers": {"memorylayer": { "url": "http://localhost:7400/sse" }}}Check it works
# Agent A stores:remember({ content: "prefers dark mode", namespace: "user" }) # Agent B recalls:recall({ query: "UI preferences", namespace: "user", preview: true })Something missing or wrong? Email support@memorylayer.in or open an issue.