Loading ToolBench
Full on-page SEO audit for any URL. Score 0-100, meta tags, headings, schema markup, images, links, load time. Prioritized fixes.
Two tools with clear, detailed descriptions and proper input schemas. Tool naming follows verb_noun convention (seo_audit_*). Descriptions are comprehensive (200+ chars) with explicit use-case guidance and anti-patterns (what NOT to use for). Input schemas are properly typed with required fields. However, output schemas are documented only in description text, not as formal JSON Schema. No error handling guidance, no parameter validation rules, and no tool annotations (readOnlyHint/idempotentHint). Descriptions are LLM-optimized but lack actionable error recovery patterns.
Use this when you need to compare SEO health across multiple pages at once (up to 10 URLs). Returns the same full audit as seo_audit_page for each URL in a single call. 1. results (array) -- each entry contains the full SEO audit (score, title, meta, headings, links, images, schema, issues) 2. summary -- average score, worst-performing URL, most common issues across all pages Example output: {"results":[{"url":"https://a.com","score":85},{"url":"https://b.com","score":62}],"summary":{"avgScore":73,"worstUrl":"https://b.com"}} Use this FOR competitor analysis, sitemap audits, or comparing landing pages side by side. Essential when benchmarking multiple pages in one workflow. Do NOT use for single URLs -- use seo_audit_page instead. Do NOT use for content extraction -- use web_scrape_batch instead.
Use this when you need to analyze a webpage's on-page SEO health. Returns a structured JSON audit with score 0-100 and prioritized recommendations. 1. score (number 0-100) -- overall SEO health score 2. title -- tag content, length, keyword presence 3. metaDescription -- content, length, truncation risk 4. canonical -- canonical URL and self-referencing check 5. headings -- H1/H2/H3 hierarchy, count, missing H1 flag 6. links -- internal count, external count, broken link flags 7. images -- total count, missing alt text count 8. schema -- Schema.org types detected (Article, Product, FAQ, etc.) 9. openGraph -- og:title, og:image, og:description completeness 10. performance -- page load time in ms, word count 11. issues -- prioritized array of problems with severity (critical/warning/info) Example output: {"score":72,"title":{"content":"My Page","length":7},"headings":{"h1":1,"h2":3},"images":{"total":5,"missingAlt":2},"issues":[{"severity":"critical","message":"Missing meta description"}]} Use this BEFORE optimizing any webpage, writing SEO recommendations, or auditing a client's site. Essential for competitive page-level analysis and content audits. Do NOT use for content extraction -- use web_scrape_to_markdown instead. Do NOT use for screenshots -- use capture_screenshot instead. Do NOT use for tech detection -- use website_detect_tech_stack instead. Do NOT use for domain-level data (WHOIS/DNS) -- use domain_lookup_intelligence instead.
Output schemas documented only in description text, not as formal JSON Schema. LLMs cannot parse structured output constraints from prose.
No error handling guidance. Descriptions lack recovery hints (e.g., 'If URL is invalid, verify it starts with https://'). Agents cannot self-correct on failures.
No tool annotations (readOnlyHint, idempotentHint). Both tools are read-only and idempotent but this is not declared in the tool definition.
seo_audit_batch accepts up to 10 URLs but no validation rules documented. No guidance on timeout, rate limits, or per-URL failure handling.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | C | 66 | 2025-06-18+ | v2 |
Parameter descriptions lack format/constraint details. 'url' param has no guidance on valid schemes, max length, or what happens with invalid URLs.