Three tools with clear verb-noun naming (get_*) and basic descriptions. However, all tools lack documented output schemas, critical for LLM planning. Parameter descriptions are minimal (1-2 sentences). No error handling guidance, no pagination support, and no indication of what fields are returned. Descriptions are adequate (60-90 chars) but lack WHEN/WHY context. Tools are read-only and simple, but schema documentation is entirely absent, which violates pattern:tool and pattern:response-shaper.

• Document output schemas for all three tools. Specify: What fields are returned? Are they nested objects or flat? What types (string, array, object)? Example: get-component-usage returns {name, description, props: [{name, type, required, description}], variants: [...], accessibility: {...}}.
• Add a list-components tool that returns {components: [{name, category, description}], total: number} with pagination (limit, offset). This enables agents to discover valid component names before calling get-component-usage.
• Enhance parameter descriptions with format constraints and error recovery. E.g. 'Component name (case-sensitive, registered in Hareru UI). If not found, call list-components to see available options.'
• Add error handling guidance in tool descriptions. E.g. 'Returns 404 if component not found. Try list-components or search-components to find the correct name.'
• Clarify the distinction between get-component-usage and get-bundle-usage in their descriptions. When should an agent call one vs the other? What is a 'bundle' vs a 'component'?