Store entities, relations and observations in a local knowledge graph with the reference Memory MCP server.
The reference Memory MCP server gives an assistant a persistent knowledge graph. It stores named entities, relationships between them and text observations attached to entities, making it useful for project facts that should be retrieved explicitly in a later task.
For a project, create an entity for a service or decision and attach a concise observation. Connect related entities with a named relationship. Ask the assistant to read or search the graph before relying on an earlier fact. Installing the server does not automatically import every past conversation.
Run @modelcontextprotocol/server-memory with npx. The server stores its graph in a JSONL file; MEMORY_FILE_PATH selects a custom location. Use a stable absolute path if you want the same graph across sessions. This is a graph and text-search reference implementation, so avoid assuming it provides a separate vector database or semantic retrieval engine.
Check the linked publisher documentation and your connected client's tool list. Names and availability can vary with the server version and account permissions.
| Tool | What it does |
|---|---|
| create_entities | Create named graph entities with observations. |
| create_relations | Connect existing entities. |
| add_observations | Add facts to entities. |
| search_nodes | Search graph nodes. |
| read_graph | Retrieve the stored knowledge graph. |
{
"mcpServers": {
"memory": {
"command": "npx",
"args": [
"-y",
"@modelcontextprotocol/server-memory"
],
"env": {
"MEMORY_FILE_PATH": "/absolute/path/to/project-memory.jsonl"
}
}
}
}Replace the example path with a writable absolute file path.
Node.js with npx and a writable location for the graph file.
| Variable | Description | Required |
|---|---|---|
| MEMORY_FILE_PATH | Optional custom path for the JSONL graph file. | Optional |
A structured scratchpad for hard problems — stepwise reasoning with revisions, branches and visible logic.
Symbol-level code navigation, refactoring and memory for coding agents — the IDE brain your assistant has been missing.
A Model Context Protocol (MCP) server for interacting with the Backlog API. This server provides tools for managing projects, issues, wiki pages, and
A Model Completion Protocol (MCP) server implementation for ServiceNow, allowing Claude to interact with ServiceNow instances.
Enable "Vibe Coding" with PixelLab's Model Context Protocol (MCP) server. Generate pixel art characters, animations, and tilesets directly from your
Condition-aware ingredient & product safety for AI agents: verdict, evidence tier, citations.