Skip to content

MCP Server Tool Reference (Auto-Generated)

Auto-Generated Documentation

Generated from MCP server tool schemas. Last updated: 2025-11-30

Overview

The Knowledge Graph MCP server provides tools for Claude Desktop to interact with the knowledge graph. These tools enable semantic search, concept exploration, and graph traversal directly from Claude.

Available Tools

  • search - Search for concepts or source passages using semantic similarity. Your ENTRY POINT to the graph.

CONCEPT SEARCH (type: "concepts", default) - Find concepts by semantic similarity: - Grounding strength (-1.0 to 1.0): Reliability/contradiction score - Diversity score: Conceptual richness (% of diverse connections) - Authenticated diversity: Support vs contradiction indicator (✅✓⚠❌) - Evidence samples: Quoted text from source documents - Image indicators: Visual evidence when available - Document sources: Where concepts originated

SOURCE SEARCH (type: "sources") - Find source text passages directly (ADR-068): - Searches source document embeddings, not concept embeddings - Returns matched text chunks with character offsets for highlighting - Shows concepts extracted from those passages - Useful for RAG workflows and finding original context

RECOMMENDED WORKFLOW: After search, use concept (action: "connect") to find HOW concepts relate - this reveals narrative flows and cause/effect chains that individual searches cannot show. Connection paths are often more valuable than isolated concepts.

Use 2-3 word phrases (e.g., "linear thinking patterns"). - concept - Work with concepts: get details (ALL evidence + relationships), find related concepts (neighborhood exploration), or discover connections (paths between concepts).

PERFORMANCE CRITICAL: For "connect" action, use threshold >= 0.75 to avoid database overload. Lower thresholds create exponentially larger searches that can hang for minutes. Start with threshold=0.8, max_hops=3, then adjust if needed. - ontology - Manage ontologies (knowledge domains/collections): list all, get info, list files, or delete. Use action parameter to specify operation. - job - Manage ingestion jobs: get status, list jobs, approve, or cancel. Use action parameter to specify operation.

Search for concepts or source passages using semantic similarity. Your ENTRY POINT to the graph.

CONCEPT SEARCH (type: "concepts", default) - Find concepts by semantic similarity: - Grounding strength (-1.0 to 1.0): Reliability/contradiction score - Diversity score: Conceptual richness (% of diverse connections) - Authenticated diversity: Support vs contradiction indicator (✅✓⚠❌) - Evidence samples: Quoted text from source documents - Image indicators: Visual evidence when available - Document sources: Where concepts originated

SOURCE SEARCH (type: "sources") - Find source text passages directly (ADR-068): - Searches source document embeddings, not concept embeddings - Returns matched text chunks with character offsets for highlighting - Shows concepts extracted from those passages - Useful for RAG workflows and finding original context

RECOMMENDED WORKFLOW: After search, use concept (action: "connect") to find HOW concepts relate - this reveals narrative flows and cause/effect chains that individual searches cannot show. Connection paths are often more valuable than isolated concepts.

Use 2-3 word phrases (e.g., "linear thinking patterns").

Parameters:

  • query (string) (required) - Search query text (2-3 word phrases work best, e.g., "linear thinking patterns")
  • type (string) - Search type: "concepts" (default - semantic concept search) or "sources" (source passage search, ADR-068)
  • Allowed values: concepts, sources
  • Default: "concepts"
  • limit (number) - Maximum number of results to return (default: 10, max: 100)
  • Default: 10
  • min_similarity (number) - Minimum similarity score 0.0-1.0 (default: 0.7 for 70%, lower to 0.5-0.6 for broader matches)
  • Default: 0.7
  • offset (number) - Number of results to skip for pagination (default: 0)
  • Default: 0
  • ontology (string) - Filter by ontology/document name (sources only)

concept

Work with concepts: get details (ALL evidence + relationships), find related concepts (neighborhood exploration), or discover connections (paths between concepts).

PERFORMANCE CRITICAL: For "connect" action, use threshold >= 0.75 to avoid database overload. Lower thresholds create exponentially larger searches that can hang for minutes. Start with threshold=0.8, max_hops=3, then adjust if needed.

Parameters:

  • action (string) (required) - Operation: "details" (get ALL evidence), "related" (explore neighborhood), "connect" (find paths)
  • Allowed values: details, related, connect
  • concept_id (string) - Concept ID (required for details, related)
  • include_grounding (boolean) - Include grounding_strength (default: true)
  • Default: true
  • include_diversity (boolean) - Include diversity metrics for details action (default: false, adds ~100-500ms)
  • Default: false
  • diversity_max_hops (number) - Max hops for diversity calculation (default: 2)
  • Default: 2
  • truncate_evidence (boolean) - Truncate evidence full_text context to 200 chars (default: true for token efficiency). Set false for complete context.
  • Default: true
  • max_depth (number) - Max traversal depth for related (1-5, default: 2)
  • Default: 2
  • relationship_types (array) - Filter relationships (e.g., ["SUPPORTS", "CONTRADICTS"])
  • include_epistemic_status (array) - Only include relationships with these epistemic statuses (e.g., ["AFFIRMATIVE", "CONTESTED"])
  • exclude_epistemic_status (array) - Exclude relationships with these epistemic statuses (e.g., ["HISTORICAL", "INSUFFICIENT_DATA"])
  • connection_mode (string) - Connection mode: "exact" (IDs) or "semantic" (phrases)
  • Allowed values: exact, semantic
  • Default: "semantic"
  • from_id (string) - Starting concept ID (for exact mode)
  • to_id (string) - Target concept ID (for exact mode)
  • from_query (string) - Starting phrase (for semantic mode, 2-3 words)
  • to_query (string) - Target phrase (for semantic mode, 2-3 words)
  • max_hops (number) - Max path length (default: 3). WARNING: Values >5 combined with threshold <0.75 can cause severe performance issues.
  • Default: 3
  • threshold (number) - Similarity threshold for semantic mode (default: 0.75). PERFORMANCE GUIDE: 0.85+ = precise/fast, 0.75-0.84 = balanced, 0.60-0.74 = exploratory/SLOW, <0.60 = DANGEROUS (can hang database for minutes)
  • Default: 0.75

ontology

Manage ontologies (knowledge domains/collections): list all, get info, list files, or delete. Use action parameter to specify operation.

Parameters:

  • action (string) (required) - Operation: "list" (all ontologies), "info" (details), "files" (source files), "delete" (remove)
  • Allowed values: list, info, files, delete
  • ontology_name (string) - Ontology name (required for info, files, delete)
  • force (boolean) - Confirm deletion (required for delete)
  • Default: false

job

Manage ingestion jobs: get status, list jobs, approve, or cancel. Use action parameter to specify operation.

Parameters:

  • action (string) (required) - Operation: "status" (get job status), "list" (list jobs), "approve" (approve job), "cancel" (cancel job)
  • Allowed values: status, list, approve, cancel
  • job_id (string) - Job ID (required for status, approve, cancel)
  • status (string) - Filter by status for list (pending, awaiting_approval, running, completed, failed)
  • limit (number) - Max jobs to return for list (default: 50)
  • Default: 50