The server defines 3 tools with basic schemas and descriptions. Naming follows verb-noun convention (list_, analyze_, scan_) which is good. However, descriptions are generic and lack depth, they do not explain WHEN to use each tool, what the outputs look like, or how they differ from each other. Parameter descriptions are minimal. Error handling returns JSON with an 'error' field, but does not guide the LLM on recovery steps or classify errors. Output schemas are not documented, it's unclear what fields list_rules returns, or what the structure of analyze_element/scan_page results look like. The server validates URLs and handles browser lifecycle, but these operational details are not reflected in tool descriptions or error guidance.

• Document the output schema for each tool. For list_rules, specify the structure: [{id: string, label: string, requiredProperties: string[]}]. For analyze_element and scan_page, document what a violation object looks like (e.g., {ruleId: string, severity: 'error'|'warning', element: string, message: string}).
• Expand tool descriptions to include WHEN to use each tool. Example: 'Call list_rules first to understand available checks. Use analyze_element to check a single element quickly; use scan_page to audit the entire page for violations. Returns a list of CSS no-op violations with rule IDs and suggested fixes.'
• Add parameter descriptions explaining constraints and failure modes. For 'url': 'URL of the page to scan (must be publicly accessible, supports http/https, will wait up to 30s for page load).' For 'selector': 'CSS selector matching the element to analyze (e.g., #main, .btn, div[data-id]). If the selector matches multiple elements, analyzes the first match.'
• Classify and guide error recovery. Instead of 'Analysis failed: ...' return specific messages: 'URL not accessible (timeout after 30s). Check the URL is valid and the server is reachable.' or 'Selector did not match any elements on the page. Try a broader selector or inspect the page structure.'
• Consider adding a 'get_rule_details' tool that returns detailed information about a specific rule (id, label, examples of violations, how to fix them). This enables multi-step workflows: list rules → get details → analyze element → review violations.