What breaks without openclaw graphiti memory
No cross-session recall. Flat memory stores. Relationship context lost.
→
Persistent graph memory × temporal knowledge graph ÷ 2-hour integration ÷ no flat-store limitations = agents that remember.
Security check — openclaw graphiti memory
Privacy score: 7/10 — accesses connected platform APIs only.
Lock it: review OAuth scopes before install, confirm OpenClaw ≥1.2; Neo4j ≥5; Node.js ≥18; Python 3.10+ for Graphiti SDK compatibility.
Quick start — openclaw graphiti memory in 2–4 hours
Setup time: 2–4 hours
!
You need:
- OpenClaw core ≥1.2
- Neo4j ≥5 or compatible graph DB
- Graphiti SDK
- embedding API
Install the package:
npm install @clawdbrunner/openclaw-graphiti-memory
# Set NEO4J_URI, NEO4J_USER, NEO4J_PASSWORD, EMBEDDING_API_KEY in .env
1
Set up Neo4j (local or cloud)
3
Configure Neo4j and embedding credentials in .env
4
Register in openclaw.config.js
5
Agents can now call memory.remember() and memory.recall()
6
Verify stored facts via the Neo4j browser
Compatibility & status
Works with: OpenClaw ≥1.2; Neo4j ≥5; Node.js ≥18; Python 3.10+ for Graphiti SDK
advanced
Last updated: Oct 2025
★ 265 on GitHub
MIT
Official docs →
View on GitHub →
FAQ — openclaw graphiti memory
How does this differ from openclaw-supermemory?
Supermemory uses a vector store for semantic similarity. Graphiti memory uses a knowledge graph for structured, relational facts.
Is Graphiti the same as the open-source Graphiti by Zep AI?
Yes — this plugin wraps the Zep AI Graphiti framework.
Can I use both in the same bot?
Yes — they serve complementary memory roles.