Kopeng has 19 tools with consistent naming (verb_noun pattern: store_memory, search_memories, get_memory, etc.) and all have descriptions. However, descriptions are brief (avg ~40 chars, well below the 194-char baseline for A+ tools) and lack context on WHEN to use each tool or dependencies between them. Input schemas are present and typed, but parameter descriptions are minimal (avg ~30 chars vs 72-char baseline). No output schemas documented. Error handling is absent from tool definitions. Tools like trigger_discovery and trigger_dream lack clarity on side effects and prerequisites. The semantic memory system is well-structured but under-documented for LLM reasoning.

• Expand all tool descriptions to 100-200 chars. Include: what it does, when to use it (vs similar tools), and what it returns. E.g., 'search_memories: Find memories by semantic similarity. Use this to retrieve relevant context before storing new memories. Returns up to [limit] memories with ID, content, scope, and relevance score.'
• Add range constraints to numeric parameters: limit (1-100, default 20), offset (0+, default 0), depth (1-5, default 2). Add enum constraints to string parameters: scope (optional, any project name), approved (true|false).
• Document output schemas for all tools. E.g., search_memories returns [{id: number, content: string, scope: string, relevance: number, tags: string[]}]. Include this in tool descriptions or a separate schema field.
• Clarify trigger_* tools: are they async? Do they return a job_id or status? Can they be called multiple times? Add descriptions like 'Triggers an async discovery process. Returns job_id. Poll list_pending_dreams to check progress.'
• Add error recovery hints to tool descriptions. E.g., 'If no memories found, try: (1) broaden the query, (2) call trigger_discovery to index new content, (3) check scope filter is correct.'

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

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