1
Add it to Claude Code
claude mcp add -e "DATABASE_URL=${DATABASE_URL}" prism-mcp -- npx -y prism-mcp-serverRequired:
DATABASE_URL+ 1 optional2
Make your agent remember this setup
prism-mcp's config, env vars, and the gotchas you hit — recalled in every future Claude Code, Cursor, and Codex session.
npx conare@latestFree · one command · indexes the sessions already on disk. Set up in the browser instead →
What it does
- Persistent memory that survives between AI agent sessions
- SQLite-based storage with F32_BLOB vector search
- GDPR-compliant memory deletion with tombstoning
- Memory tracing with latency breakdown for observability
- LangChain and LangGraph integration for agentic workflows
Tools 1
session_forget_memoryDeletes memory associated with a session using soft or hard delete methods.Environment Variables
DATABASE_URLrequiredPath to the SQLite database file for persistent storage.PRISM_DEBUG_LOGGINGEnables verbose startup logs for debugging.Try it
→Recall the project requirements we discussed in our last session.
→Forget all memory associated with the previous testing session for GDPR compliance.
→Search my memory for the research notes I saved regarding the new API integration.
→Provide a summary of the key decisions made during our last three development sessions.
Original README from dcostenco/prism-mcp
Prism MCP — The Mind Palace for AI Agents 🧠
Your AI agent's memory that survives between sessions. Prism MCP is a Model Context Protocol server that gives Claude Desktop, Cursor, Windsurf, and any MCP client persistent memory, time travel, visual context, multi-agent sync, GDPR-compliant deletion, memory tracing, and LangChain integration — all running locally with zero cloud dependencies.
Built with SQLite + F32_BLOB vector search, optimistic concurrency control, MCP Prompts & Resources, auto-compaction, Gemini-powered Morning Briefings, MemoryTrace explainability, and optional Supabase cloud sync.
What's New in v2.5.0 — Enterprise Memory 🏗️
| Feature | Description |
|---|---|
| 🔍 Memory Tracing (Phase 1) | Every search now returns a structured MemoryTrace with latency breakdown (embedding_ms, storage_ms, total_ms), search strategy, and scoring metadata — surfaced as a separate content[1] block for LangSmith integration. |
| 🛡️ GDPR Memory Deletion (Phase 2) | New session_forget_memory tool with soft-delete (tombstoning via deleted_at) and hard-delete. Ownership guards prevent cross-user deletion. deleted_reason column captures GDPR Article 17 justification. Top-K Hole solved by filtering inside SQL, not post-query (ensures we always return exactly K results, rather than returning fewer because deleted items were filtered out after the vector search). |
| 🔗 LangChain Integration (Phase 3) | PrismMemoryRetriever and PrismKnowledgeRetriever — async-first BaseRetriever subclasses that wrap Prism MCP's traced search endpoints. Trace metadata flows automatically into Document.metadata["trace"] for LangSmith visibility. |
| 🧩 LangGraph Research Agent | Full example in examples/langgraph-agent/ — a 5-node agentic research loop with MCP bridge, persistent memory, and EnsembleRetriever hybrid search. |
| Feature | Description |
|---|---|
| 🔄 Dynamic Versioning | Server version is now derived from package.json at startup — MCP handshake, dashboard badge, and npm metadata always stay in sync. Falls back to 0.0.0 if unreadable. |
| 🛡️ Embedding Dimension Validation | generateEmbedding() now validates the returned vector is exactly 768 dimensions at runtime, catching model regressions before storing bad vectors. Removed as any cast in favor of proper EmbedContentRequest typing. |
| Feature | Description |
|---|---|
| 🪲 Windows Black Screen Fix | Fixed Python subprocess.Popen spawning visible Node.js terminal windows on Windows. |
| 📝 Debug Logging | Gated verbose startup logs behind PRISM_DEBUG_LOGGING for a cleaner default experience. |
| ⚡ Excess Loading Fixes | Performanc |