google-seo-auditlisted

This skill should be used when the user asks to "audit my page/site/repo", "scan for SEO", "check structured data / rich results", "is my page Google-ready", "SEO audit", "Google scanner", or wants prioritized, cited recommendations to improve a page in Google Search. It runs a full multi-agent SEO / discoverability audit of a web page - a live URL OR a local .html file in the current repo - grounded in the local Google documentation knowledge base (the google-search-ads-analytics-docs skill): it spawns one subagent per audit dimension and produces a scored report with fixes that cite the official Google docs (source_url).
bsisduck/google-search-ads-analytics-docs · ★ 0 · Data & Documents · score 65

Install: claude install-skill bsisduck/google-search-ads-analytics-docs

# Google SEO Audit - multi-agent page audit Audits a page across every Google-discoverability dimension and returns a **scored report with prioritized, cited recommendations**. Every recommendation must be grounded in the local corpus via the **`google-search-ads-analytics-docs`** skill (`.claude/skills/google-search-ads-analytics-docs/search.py`) and cite a Google `source_url`. ## Fast path - the Workflow For a thorough, adversarially-verified audit, run the bundled **`seo-audit`** workflow (Workflow tool), saved at `.claude/workflows/seo-audit.js`. It implements the four phases below and **scales agents to the work - no fixed cap**: **Snapshot -> Verify** (one agent per dimension) **-> Refute** (adversarial verifiers per finding, false positives dropped) **-> Synthesize** (`AUDIT-<page>.md`). Pass the target as args (URL or local path); optional `{ refuters: N }` raises the adversarial panel size. If the Workflow tool isn't available, do the same steps manually with the Task tool, below. ## Step 1 - Pick the target(s) - A live page: pass the URL. - **This repo (default when the user says "my page/repo"):** find local pages with `Glob "**/*.html"` (ignore `.venv/`, `node_modules/`). The repo's page is `app/index.html`. Audit each found page. ## Step 2 - Snapshot (deterministic, run once per target) Use a per-target output file so multiple pages don't clobber each other: ```bash .venv/bin/python3 .claude/skills/google-seo-audit/fetch_page.py <url-or-path> --out /tmp/