# StudioMeyer Memory

> Long-term AI memory with bi-temporal knowledge graph. Provider-agnostic memory backend that works with Claude, GPT, Gemini, or any LLM speaking MCP. 56 tools, multi-tenant on our own Postgres at Hetzner Nuremberg (Germany, EU), with tenant-scoped application queries.

## What this server does

Persists memories that survive across conversations, sessions, and AI clients. Stores learnings, decisions, sessions, skills, and a knowledge graph of entities + observations + relations. Search uses trigram + full-text + temporal decay. Bi-temporal model supports time-travel queries (what did we know on date X?).

## Provider-agnostic

Use any LLM through an MCP-capable client or a REST integration. Your client or workflow chooses the model; Memory provides storage and search. Importing ChatGPT, Claude or Gemini history does not imply a direct connection to those chat products.

## Connect

- **MCP endpoint**: https://memory.studiomeyer.io/mcp (JSON-RPC 2.0 over HTTP, Bearer auth)
- **Service discovery**: https://memory.studiomeyer.io/.well-known/agents.json (MCP metadata)
- **OAuth 2.1**: https://memory.studiomeyer.io/.well-known/oauth-authorization-server
- **MCP protected resource metadata**: https://memory.studiomeyer.io/.well-known/oauth-protected-resource/mcp
- **REST OpenAPI**: https://memory.studiomeyer.io/openapi.json (documented data API subset)
- **OAuth scope**: memory (Full access to the memory operations available to your account); separate read/write permissions are not implemented
- **Documentation**: https://studiomeyer.io/services/memory

## Get an API key

Sign in to https://studiomeyer.io/portal/login then click "Free Memory testen". The Free plan includes 5,000 API calls/day, with stored-data limits. No credit card required.

## Top capabilities

- nex_search — unified RRF search across decisions, learnings, sessions, entities
- nex_learn — store typed learning with duplicate-prevention gatekeeper
- nex_decide — log decision with confidence + alternatives
- nex_entity_create / observe / search / merge — knowledge graph CRUD
- nex_session_start / end — auto-loaded context + structured summaries
- nex_skill — track which patterns work for which domains
- nex_import — bulk-import ChatGPT / Claude / Gemini conversation history
- nex_synthesize — extract patterns from learning clusters
- nex_proactive — get suggestions based on recent activity

## Pricing

- Free: 0 EUR/month; 5,000 API calls/day; 1,000 learnings; 100 entities; 500 observations; 1 API key
- Pro: 9 EUR/month; 50,000 API calls/day; 25,000 learnings; 1,000 entities; 10,000 observations; 3 API keys
- Team: 19 EUR/month; unlimited API calls/day; unlimited learnings; unlimited entities; unlimited observations; 20 API keys

Free requires no credit card. AppSumo lifetime licenses use separate tier limits. Daily API caps are separate from the tenant request burst limit (default 60 MCP HTTP requests/minute, configurable by the service operator). Stored-data limits count saved records, not searches.

## Data handling

Memory storage is on our own PostgreSQL at Hetzner Nuremberg (Germany, EU). Isolation uses tenant-scoped application queries. AI features may send relevant content to configured model providers; EU storage does not imply EU-only model processing. Secret scanning detects supported sensitive patterns, not general PII.

Export your knowledge with nex_export; delete graph records with nex_entity_delete. Privacy details: https://studiomeyer.io/datenschutz

## Open source

- n8n custom node: https://www.npmjs.com/package/n8n-nodes-studiomeyer-memory
- MCP protocol conformance test harness: https://www.npmjs.com/package/mcp-protocol-conformance
- MCP server attestation lib: https://www.npmjs.com/package/mcp-server-attestation
- MCP hook conformance: https://www.npmjs.com/package/mcp-hook-conformance
- MCP tenant-pair: https://www.npmjs.com/package/mcp-tenant-pair

Server version: 3.47.0