memorymcp defines 16 tools with complete input schemas and descriptions. Naming follows verb_noun convention (query_memory, store_fact, add_episode). Descriptions are present but vary in quality, some are concise and actionable (e.g., query_memory at 127 chars), others are vague (memory_config at 67 chars). All parameters have type definitions and most have descriptions, but several lack guidance on constraints, valid ranges, or when to use them. Output schemas are not documented, LLMs cannot predict response structure. Error handling is absent from descriptions. No tool annotations (readOnlyHint, destructiveHint) despite clear risk levels (WRITE, DESTRUCTIVE). Parameter descriptions often omit format/constraint details (e.g., 'importance' score lacks range; 'max_hops' lacks bounds). The server targets semantic memory patterns well but lacks production-grade error recovery guidance.

• Document output schemas for all 16 tools. For query_memory, specify: {results: [{fact_id, content, score, created_at, tags}], total_count}. For store_fact, specify: {fact_id, namespace, stored_at}. This enables LLMs to plan downstream calls.
• Add tool annotations via inputSchema or metadata. Mark forget_fact with destructiveHint=true, extract_facts/store_fact/add_episode/set_working_memory/import_memory with readOnlyHint=false, and query_memory/get_session_context/get_working_memory/traverse_graph/find_contradictions/export_memory/memory_stats/memory_config/memory_analytics with readOnlyHint=true.
• Specify parameter constraints in descriptions. E.g., 'importance: 0.0 - 1.0 float indicating fact weight'; 'max_hops: 1 - 10 integer'; 'max_tokens: 100 - 8000 integer'. Add min/max to schema if JSON Schema supports it.
• Add error recovery guidance to descriptions. E.g., query_memory: 'Returns empty results if no matches found, try broader topic or check namespace. If namespace does not exist, call memory_config to list available namespaces.' forget_fact: 'Irreversible. Verify fact_id before calling.'
• Expand utility tool descriptions. memory_config: 'Call first to discover available backends (episodic, semantic, graph) and current configuration. Returns structure needed for namespace selection.' memory_stats: 'Returns fact count, memory usage, decay metrics. Use to monitor memory health before large imports.'

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

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