OpenMemoryBetter memory for your agents
OpenMemory learns your patterns, preferences, and decisions — then serves them to any AI agent automatically. Works with OpenClaw, Cursor, Claude, VS Code, Codex, and any MCP-compatible tool.
Install once. Memory handles itself.
OpenMemory captures context as you work, organizes it by project, and serves it to your agents when they need it.
async def verify_token(request: Request):
token = request.headers.get("Authorization",
"").replace("Bearer ", "")
try:
payload = jwt.decode(token, SECRET_KEY,
algorithms=["HS256"])
return payload.get("user_id")
except JWTError:
raise HTTPException(
status_code=401,
detail="Invalid token")Capture
Saves decisions, patterns, and preferences as you work. No manual tagging needed.
Projects Overview
Global Preferences
Organize
Memories are scoped by project and tagged by type. Browse, edit, or delete anything.
Based on your stored memories, JWT authentication in this project uses PyJWT with HS256 signing. Tokens are passed via the Authorization header, decoded with jwt.decode(), and validated against a shared SECRET_KEY. Failed tokens return a 401 HTTPException.
Deliver
When an agent starts a task, it pulls relevant memories. No re-explaining.
Stop re-explaining your codebase
Your coding agent reviews the same patterns, breaks the same conventions, and forgets your feedback every session. OpenMemory fixes that.
- Learns your team's coding patterns and conventions
- Remembers past code review feedback
- Carries context across files, sessions, and projects
- Fewer review cycles. Better first drafts.
Using your team's convention: error responses use ProblemDetails format with traceId.
I remember you prefer morning meetings and async standups. I've scheduled the sync for 9 AM and set up a thread for updates.
Give OpenClaw and your personal agents long-term memory
Personal AI agents like OpenClaw start from zero every conversation. OpenMemory gives them persistent memory — your preferences, past decisions, and context carry over automatically.
- Give OpenClaw agents memory that persists across every session
- Remembers your preferences, routines, and recurring tasks
- Carries context across tools — OpenClaw, Claude, ChatGPT, and more
- You control what's stored and who can access it
Tagged and searchable memory
Every memory is tagged by type — preferences, decisions, patterns, troubleshooting — so your agent pulls the right context every time.
Scoped to the right project
Memories are organized by project. When you switch repos or tasks, your agent automatically gets the relevant context.
Full visibility and control
See every memory that was added, read, or edited — and by which agent. Edit, delete, or restrict access to any memory at any time.
JWT authentication middleware for FastAPI — extracts Bearer token from Authorization header, decodes using jose library with HS256 algorithm, returns user_id from payload, raises 401 HTTPException on invalid tokens
Claude12 Min agoSEARCH
Cursor12 Apr 2025, 8:30 AMSEARCH
Codex11 Apr 2025, 10:12 PMSEARCHTop Organizations Building with OpenMemory
Top teams using OpenMemory
Frequently Asked Questions
Your agents should remember.
Install OpenMemory and stop repeating yourself.