Single tool with basic schema and description. Tool name 'query_strands_docs' is action-oriented but generic. Description (95 chars) is present but minimal, lacks WHEN to use, prerequisites, or output structure guidance. Input schema properly typed with required field, but no output schema documented. No error handling guidance, no parameter constraints, no pagination support despite knowledge base queries likely returning multiple results. Phase-2 implementation adds AWS error handling but still lacks structured output documentation and recovery guidance.

• Add explicit output schema to tool definition: document that response includes 'answer' (string), 'sources' (array of {uri, title}), and 'confidence' (0-1). This lets LLM plan follow-up actions.
• Expand description to 150-200 chars: 'Query AWS Bedrock Knowledge Base for Strands Agent documentation. Returns generated answer with source citations. Requires AWS credentials and bedrock:RetrieveAndGenerate IAM permission. Use when user asks about Strands features, configuration, or troubleshooting.'
• Add query parameter constraints: minLength 3, maxLength 500, pattern for alphanumeric + spaces/punctuation. Include in description: 'Query string (3-500 chars, alphanumeric + punctuation).'
• Implement pagination: add optional 'max_results' (1-50, default 20) and 'next_token' parameters. Return response with 'has_more' boolean and 'next_token' for continuation.
• Enhance error responses with recovery guidance: instead of 'AWS error (AccessDeniedException): ...', return 'Permission denied. Ensure your AWS user has bedrock:RetrieveAndGenerate permission. Contact your AWS admin or check IAM policy.'