Wisdomgraph
Persist DIKW-structured memory in Neo4j so Claude Code sessions can recall and compound project knowledge over time.
Overview
wisdomGraph is a MCP server for the Build phase that stores Neo4j-backed DIKW memory so Claude Code agents retain and compound knowledge across sessions.
What is this MCP server?
- Neo4j-native DIKW (Data–Information–Knowledge–Wisdom) memory model for agents
- stdio npm package wisdomgraph version 0.2.0
- Designed for Claude Code with knowledge that compounds across sessions
- Requires WISDOM_NEO4J_PASSWORD via wisdom connect or environment export
- Package version 0.2.0 with stdio transport
- One required secret environment variable: WISDOM_NEO4J_PASSWORD
- Source repository: github.com/cklam12345/wisdomGraph (npm subfolder)
Community signal: 2 GitHub stars.
What problem does it solve?
Agent chats reset context, so solo builders repeat explanations and rediscover the same architectural facts every session.
Who is it for?
Builders already comfortable with Neo4j who want durable, relational agent memory on their machine.
Skip if: Quick prototypes with no database appetite or teams that only need ephemeral chat history.
What do I get? / Deliverables
Sessions read and write structured graph memory so decisions and learnings persist and strengthen over time.
- Graph-stored facts and relationships accessible from the agent via MCP tools
- Cross-session recall of project knowledge instead of one-off chat context
- Compounding DIKW layers as you document more implementation work
Recommended MCP Servers
Journey fit
How it compares
Neo4j memory MCP, not a hosted SaaS knowledge wiki or flat vector note app.
Common Questions / FAQ
Who is wisdomGraph for?
Claude Code users and other MCP clients who want graph-shaped, session-spanning project memory on their own Neo4j database.
When should I use wisdomGraph?
Use it once a codebase or product has enough moving parts that you need persisted entities, relationships, and distilled wisdom between agent runs.
How do I add wisdomGraph to my agent?
Install the npm wisdomgraph package, configure Neo4j and WISDOM_NEO4J_PASSWORD, register the stdio server in your MCP config, then use wisdom connect if you follow the upstream workflow.