Single tool 'knowledge_search' has a basic description (54 chars) and minimal input schema. Description lacks WHEN to use it, prerequisites, or return structure. Input schema defines 'query' and 'limit' with types and brief descriptions, but no output schema is documented. No error handling guidance, no pagination details despite 'limit' parameter suggesting result sets. Tool name is acceptable (verb_noun form) but description is too terse to guide LLM selection effectively. No evidence of idempotency, security scoping, or composition patterns.

• Expand description to 100 - 200 chars: 'Search the knowledge base for documents matching a query. Returns ranked results with relevance scores. Use this to find relevant documentation, FAQs, or archived articles before answering user questions. Returns up to [limit] results; use pagination for large result sets.'
• Document output schema explicitly: 'Returns {results: [{id, title, content, relevance_score, url}], total_count, has_more}'
• Add pagination details: 'limit parameter caps results at 50; if total_count > limit, use offset parameter in next call to fetch additional pages.'
• Add error handling: 'If no results found, return empty results array with suggestion: "No documents matched. Try broader keywords or check spelling."'
• Add tool annotation: mark as readOnlyHint=true (search does not modify state) so LLM knows it is safe to retry.