exploring-knowledge-graph
Guidance for deep knowledge graph traversal across memories, entities, and relationships. Use when needing comprehensive context before planning, investigating connections between concepts, or answering "what do you know about X" questions.
When & Why to Use This Skill
This Claude skill enables deep knowledge graph traversal to provide comprehensive context for AI agents. By navigating interconnected memories, entities, and relationships, it solves the problem of fragmented information retrieval, allowing for more informed planning, complex concept mapping, and thorough answers to broad knowledge-based queries beyond simple keyword searches.
Use Cases
- Complex Project Planning: Understanding historical decisions, patterns, and existing documentation across multiple topics before initiating new work.
- Comprehensive Knowledge Discovery: Answering broad questions like 'What do you know about X?' by retrieving direct matches, linked memories, and related entities.
- Relationship Mapping: Investigating how different concepts, projects, and people connect within a large organizational knowledge base to identify hidden dependencies.
- Contextual Onboarding: Quickly gathering all relevant background information, code artifacts, and document links for a specific topic to ensure high-quality, informed agent output.
| name | exploring-knowledge-graph |
|---|---|
| description | Guidance for deep knowledge graph traversal across memories, entities, and relationships. Use when needing comprehensive context before planning, investigating connections between concepts, or answering "what do you know about X" questions. |
| license | MIT |
| version | 1.0.0 |
| model | claude-sonnet-4-5 |
Exploring the Knowledge Graph
Forgetful stores knowledge as an interconnected graph: memories link to other memories, entities link to memories, and entities relate to each other. Deep exploration reveals context that simple queries miss.
When to Explore
Explore the knowledge graph when:
- Starting complex work that spans multiple topics
- User asks "what do you know about X"
- Planning requires understanding existing decisions/patterns
- Investigating how concepts connect across projects
- Need comprehensive context, not just top search results
Exploration Phases
Track visited IDs to prevent cycles. Execute phases sequentially.
Phase 1: Semantic Entry Point
execute_forgetful_tool("query_memory", {
"query": "<topic>",
"query_context": "Exploring knowledge graph for comprehensive context",
"k": 5,
"include_links": true,
"max_links_per_primary": 5
})
Collect: primary_memories + linked_memories (1-hop connections).
Phase 2: Expand Memory Details
For key memories, get full details:
execute_forgetful_tool("get_memory", {"memory_id": <id>})
Extract: document_ids, code_artifact_ids, project_ids, additional linked_memory_ids.
Phase 3: Entity Discovery
Find entities in discovered projects:
execute_forgetful_tool("list_entities", {
"project_ids": [<discovered project ids>]
})
Phase 4: Entity Relationships
For relevant entities, map relationship graph:
execute_forgetful_tool("get_entity_relationships", {
"entity_id": <id>,
"direction": "both"
})
Relationship types: works_for, owns, manages, collaborates_with, etc.
Phase 5: Entity-Linked Memories
For each entity, find all linked memories:
execute_forgetful_tool("get_entity_memories", {
"entity_id": <id>
})
Returns {"memory_ids": [...], "count": N}. Fetch any new memories not already visited.
Presenting Results
Group findings by type:
Memories: Primary (direct matches) → Linked (1-hop) → Entity-linked (via entities)
Entities: Name, type, relationship count, linked memory count
Artifacts: Documents and code snippets found via memory links
Graph Summary: Total nodes, key themes, suggested follow-up queries
Depth Control
- Shallow (phases 1-2): Quick context, ~5-15 memories
- Medium (phases 1-4): Include entities and relationships
- Deep (all phases): Full graph traversal, comprehensive context
Match depth to task complexity. Start shallow, go deeper if context insufficient.
Efficiency Tips
- Check
truncatedflag from query_memory (8000 token budget) - Skip Phase 3-5 if no entities exist in discovered projects
- Use
project_idsfilter to scope exploration - Stop expanding when hitting diminishing returns