Graph Memory defines 3 tools with basic schemas and descriptions, but falls short of production quality. Tool names lack clear action verbs (gm_search, gm_record, gm_stats use prefixes rather than verb_noun patterns). Descriptions are present but generic (10-80 chars, below the 50-200 char LLM-optimized baseline). Parameter descriptions exist but lack detail on constraints, formats, and valid ranges. Output schemas are not documented, LLMs cannot infer what fields to expect from responses. Error handling is absent from visible code. The codebase shows sophisticated graph logic (PageRank, community detection, vector dedup) but this complexity is not reflected in tool definitions that would help LLMs reason about capabilities.

• Rename tools to verb_noun pattern: gm_search → search_knowledge_graph, gm_record → record_knowledge_node, gm_stats → get_graph_statistics. This helps LLMs parse intent from names alone.
• Document output schemas explicitly. For gm_search, specify: returns array of {node_id, node_name, node_type, description, relevance_score, edges: [{from_id, to_id, relationship_type}]}. For gm_record, specify: returns {node_id, created_at, status}.
• Convert parameter descriptions to formal constraints. Change 'type' description to enum: ['TASK', 'SKILL', 'EVENT']. Add 'limit' range: 1-100 (default 20). Add 'query' format: 'free-text search string, 1-500 characters'.
• Add error handling guidance. E.g., gm_search: 'If no results found, try a shorter query or use gm_stats to explore available nodes.' gm_record: 'If node already exists, call gm_search to retrieve its ID and update via a future update_node tool.'
• Expand tool descriptions to 100-150 chars with use-case context. E.g., gm_search: 'Search the knowledge graph by query text to find relevant nodes and relationships. Use this to recall prior context before recording new nodes. Returns up to [limit] results ranked by relevance.'

Inferred effective spec: <=2025-11-25.

• Relies on Logging (deprecated) - log to stderr or use OpenTelemetry