memory_graph

When to use

• You want to see what the memory links to a concept such as "postgresql".
• You want to check which rules or skills are attached to a technology.
• You are debugging why associative recall surfaced a result.

Parameters

Name	Type	Default	Description
noderequired	string	—	Node name or ID to explore
depth	integer	2	Traversal depth (1-3)
types	string[]	—	Filter by node types (rule, skill, concept, etc.)

Example

Arguments

{
  "node": "postgresql",
  "depth": 1,
  "types": [
    "rule",
    "skill"
  ]
}

Result shape

{
  "center": {
    "id": "n_7c1a",
    "type": "technology",
    "name": "postgresql",
    "importance": 0.72
  },
  "nodes": [
    {
      "id": "n_7c1a",
      "type": "technology",
      "name": "postgresql",
      "importance": 0.72
    },
    {
      "id": "n_91be",
      "type": "rule",
      "name": "Use CONCURRENTLY for index migrations",
      "importance": 0.55
    }
  ],
  "edges": [
    {
      "id": "e_3f20",
      "source_id": "n_91be",
      "target_id": "n_7c1a",
      "relation_type": "applies_to",
      "weight": 1
    }
  ]
}

node is looked up by name first, then by id. Returns {"error": "Node not found: ..."} if neither matches.

Values are illustrative; the keys follow the server's handler. MCP clients receive the result as JSON text content.

Server description

The description the server sends to your agent in tools/list, captured from the v14.7.0 source:

Query the unified knowledge graph. Returns neighborhood of a node: connected rules, skills, memories, concepts, entities.

Related tools

• memory_concepts
• memory_graph_stats
• memory_associate
• memory_graph_index