Run Full-Site SEO Audits With AI Agents

Runs comprehensive SEO audits with 7-15 specialist agents analyzing your site in parallel instead of sequential tools that take hours. Every recommendation carries a falsifiability check and is grounded in Google's own guidance, so you get actionable findings instead of generic best-practice lists.

1. Invocation: run /seo audit <url> to auto-detect your business type and spawn 7+ specialist agents in parallel across technical, content, schema, performance, and AI search. 2. Parallel analysis: agents run simultaneously (15 max), each analyzing their domain independently. 3. Synthesis: results aggregate into an SEO Health Score (0-100), a prioritized action plan, and optional PDF report. Output: a markdown report with critical/high/medium/low findings, each with a first-principle observation, dependency graph, falsifiability check, and leading indicator.

Full-site audits complete in 10-15 minutes instead of hours, with every recommendation falsifiable so clients can verify impact before implementation.

Requires Python 3.10+ and Claude Code CLI. Playwright Chromium (auto-installed) handles SPA rendering via --render auto auto-detection. Google APIs are optional (Tier 0-3 credentials in ~/.config/claude-seo/); the core plugin works fully offline. MCP extensions (DataForSEO, Ahrefs, SE Ranking, Profound, Bing Webmaster, Unlighthouse, Banana, Firecrawl) are opt-in and require their respective accounts but are not required for core functionality.

SPA rendering (pages with scroll-bound or post-interaction content hydration) may produce noisy findings; use the included /seo visual cross-check workflow on edge cases. Local audits without DataForSEO MCP lack competitive geo-grid and live SERP overlap data. Drift monitoring requires baseline capture before change detection works.