Three tools with basic schemas and descriptions, but significant gaps in parameter documentation and output schema clarity. Tool names lack action verbs (weaviate-store-memory vs store_memory). Descriptions are present but generic (10-50 chars for some). No parameter descriptions for 'information' and 'query' fields. No documented output schemas. No error handling guidance. No tool annotations (readOnlyHint/destructiveHint). Naming convention uses hyphens instead of underscores, deviating from standard verb_noun patterns.

• Rename tools to verb_noun format: 'store_memory', 'find_memories', 'search_knowledge_base'. Use underscores, not hyphens. This matches LLM expectations and improves discoverability.
• Add descriptions to all parameters. For 'information': 'The memory content to store (string, any length)'. For 'query': 'Natural language search query (e.g., "user's favorite color")'. Descriptions must be 20+ chars and explain the expected format.
• Document output schemas in tool descriptions. E.g., 'Returns: list of memory objects with fields: id (string), content (string), timestamp (ISO 8601)'. Include this in the description or as a separate schema field.
• Add tool annotations: mark weaviate-store-memory with destructiveHint=true; mark find/search tools with readOnlyHint=true. This enables agents to reason about side effects.
• Improve error messages with recovery guidance. Instead of 'Missing required argument "query"', return 'Missing required argument "query". Example: search_knowledge(query="What is machine learning?")'. Include the tool name and a valid example.

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

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