Recursive Enhanced Memory by Enhanced Recall
A shared, team-scoped memory system for developer sessions. Stores collective knowledge and makes it available across team members' AI assistant sessions.
REMEMBER provides a shared memory layer that allows teams to:
- Store and retrieve collective knowledge
- Maintain context across sessions and projects
- Share insights and learnings across the team
- Build a persistent organizational memory
- Team-scoped memory — share knowledge across your team
- Ownership-based truth — creators own their memories, others can confirm/refute
- Staleness surfacing — automatically flags outdated memories
- Multiple auth providers — GitHub, Google, Microsoft, Tailscale, Keycloak, Authentik, Dex, API keys, and dev mode
- Kubernetes-native — deploy anywhere with Helm or kubectl
- Container-first — Podman and Docker support
- Extensible — pluggable auth, search, and notification layers
A sci-fi themed web interface is included for browsing and managing memories without an AI assistant.
System documentation is organized as an OKF knowledge bundle — a structured collection of reference documents covering server architecture, all authentication providers, the full MCP tool surface, database schema, deployment options, and the web UI. Read okf/index.md for a complete walkthrough of how the system works and where to find details on any component.
Developer workstations Kubernetes Cluster
┌─────────────────────┐ ┌──────────────────────────┐
│ AI Assistant │ │ namespace: remember │
│ + memory plugin │ │ │
│ + CLI tool │──MCP/HTTPS──▶│ ┌────────────────────┐ │
└─────────────────────┘ │ │ remember-server │ │
│ │ (FastMCP, Python) │ │
│ │ + webui sidecar │ │
│ │ N replicas, state- │ │
│ │ less │ │
│ └──────┬───────┬─────┘ │
│ │ │ │
│ MCP │ SQL │ / │
│ (tools) │ │ (web) │
│ ▼ ▼ │
│ ┌────────────────────────┐│
│ │ remember-db ││
│ │ Postgres + pgvector ││
│ └────────────────────────┘│
└──────────────────────────┘
| Component | Description |
|---|---|
| AI Assistant | MCP client (Claude Code, etc.) connecting via Model Context Protocol |
| CLI | Local import/export and management tool |
| Web UI | Sci-fi themed FastAPI interface on port 3000 (sidecar or separate) |
| remember-server | Stateless FastMCP server, horizontally scalable |
| remember-db | Postgres with pgvector for semantic search |
Pluggable auth providers (configured via env vars or YAML):
- GitHub OAuth, Google OAuth, Microsoft/Entra ID
- Tailscale identity, Keycloak, Authentik, Dex
- API keys, dev mode
| Port | Service |
|---|---|
| 8000 | MCP server (FastMCP) |
| 3000 | Web UI (FastAPI) |
| 9090 | Prometheus metrics |
See the design docs for architecture details and the deployment guide for setup instructions.
See CONTRIBUTING.md for guidelines.
MIT
