Single tool with basic schema and description. Tool name 'get-usage' follows verb_noun pattern but description lacks depth on when/why to use it. Input schema present with Zod types but parameters lack actionable constraints (no enums, ranges, or format guidance). Output is unstructured text rather than typed fields. No error handling guidance, no pagination support despite potential for large result sets. Missing parameter descriptions explaining what vendors/products filtering actually does.

• Add parameter descriptions: 'vendors: List of cloud provider names (e.g. AWS, Azure, GCP). If omitted, returns all vendors.' and 'products: List of product/service names (e.g. EC2, RDS, VM). If omitted, returns all products.'
• Expand tool description to 100-150 chars: 'Retrieve cloud usage and cost summary across vendors and products. Call this to see resource consumption by provider and service type. Returns structured usage data with counts and totals.'
• Change output from text to structured JSON: return {vendors: [{name, rsrcType, results: [{label, total, items: [{value, count}]}]}]} so LLM can parse and chain results.
• Add pagination: accept optional 'limit' (default 50, max 500) and 'offset' parameters. Return total_count so agent knows if more results exist.
• Improve error handling: return {error: 'Failed to fetch usage data from backend API at http://localhost:8000/api/v1/assets/usage. Verify the backend service is running and accessible. Retry in 5 seconds.'} with explicit recovery steps.