Memento is an MCP server that provides persistent memory for AI tools, allowing them to retain context and knowledge across sessions without requiring cloud services or API keys. It builds a knowledge graph from conversations, enabling AI to recall information seamlessly.
From the registry: Persistent memory for AI tools. Local-first knowledge graph with hybrid search.

$ git clone https://github.com/scrypster/memento.git && cd memento && ./launch.shPlease install the `memento` MCP server into my current AI client (that's you).
Required prerequisites (do these first if not already done):
- **Docker or Go+Node+Ollama** — Docker (recommended) or Go 1.23+, Node.js 18+, Ollama (https://docker.com)
- **Clone and launch** — Clone the repo and run launch script which builds binary Run: `git clone https://github.com/scrypster/memento.git && cd memento && ./launch.sh`
Canonical MCP server config (stdio transport):
- command: `memento-mcp`
- args: []
- optional environment variables:
- `MEMENTO_DATA_PATH`: SQLite database directory (example: `./data`)
- `MEMENTO_STORAGE_ENGINE`: Storage engine: sqlite or postgres (example: `sqlite`)
- `MEMENTO_LLM_PROVIDER`: LLM provider (example: `ollama`)
- `MEMENTO_OLLAMA_URL`: Ollama API endpoint (example: `http://localhost:11434`)
- `MEMENTO_OLLAMA_MODEL`: Extraction model (example: `qwen2.5:7b`)
- `MEMENTO_OPENAI_API_KEY`: OpenAI API key (example: `<your-api-key>`)
- `MEMENTO_ANTHROPIC_API_KEY`: Anthropic API key (example: `<your-api-key>`)
- `MEMENTO_USER`: User identifier for team setups (example: `alice`)
Note: Build from source via launch.sh. Binary path from pwd/memento-mcp.
Add this MCP server to my current client's config in the correct format for you. If you need secrets or credentials I haven't provided, ASK me — do not invent values or leave raw placeholders. After adding it, tell me how to verify the server is connected.MEMENTO_USERrequiredSets the user for memory attribution.AI Agents Framework with Self Reflection and MCP support