Two tools with basic descriptions and schemas, but significant gaps in parameter documentation, error handling, and output schema clarity. index_corpus has a detailed description (good), but ask_corpus is minimal (27 chars). Neither tool documents return structure or error recovery paths. Parameters lack validation constraints (e.g., path format, question length). No error handling guidance for LLM recovery. Schemas present but incomplete, ask_corpus has no output schema documented. Tool composition is reasonable (separate index/query), but missing idempotency guarantees and rate-limit guidance.

• Expand ask_corpus description to 100 - 200 chars: 'Query the indexed document corpus with a natural-language question. Returns a summary answer with source references. Call after index_corpus completes. Requires corpus to be indexed first.'
• Document output schemas explicitly: index_corpus → {status, job_id?, indexed_files, count}; ask_corpus → {answer, sources: [{title, excerpt}], confidence}.
• Add parameter constraints: path must be absolute or relative to default storage; question must be 10 - 2000 chars. Include in descriptions: 'Path to folder (absolute or relative to default storage location).'
• Add error handling: 'If corpus is empty, index_corpus must be called first. If question is unanswerable, ask_corpus returns {answer: null, error: "No relevant documents found"}. Suggest calling index_corpus with more documents.'
• For background indexing, add a get_index_status(job_id) tool or document polling via logs. Enable LLM to verify indexing completion before calling ask_corpus.