Skip to content
Founder Workflows
BrowseSaved
Browse by stage

An open-source SEO audit agent that reads your pages with a browser and Claude for cents per URL

A free Python repo that visits each URL in a real browser, uses Claude to pull out titles, meta descriptions, headings and canonicals, checks internal links, flags pages that need a human, and writes a resumable report. Extra modules read Search Console exports.

DocumentedContent marketingSmall SaaSContent and newsletterDigital productScript

Evidence: The source shows its work: steps, screenshots, code or data you can inspect.

The business problem

A solo founder with a content or product site needs regular SEO checks (broken links, weak titles, cannibalised pages, missed quick wins) but may not want to pay for an SEO suite or audit pages by hand.

What was tried

1) List the pages to check in input.csv. 2) The core audit opens each page in a visible Chromium browser through Browser Use and Playwright. 3) The Claude API extracts SEO signals (title, meta description, H1s, canonical tag) and can suggest rewrites. 4) Links on the same domain are checked for breakage in parallel. 5) Edge cases such as login walls and 404s pause the run in interactive mode, or are logged to a 'needs human' list when run with --auto. 6) Results are written to report.json as the run goes, so an interrupted run can resume, and a plain-text summary is produced. Separate modules parse a Search Console export for quick wins and cannibalisation, score internal link sources and backlinks, map site clusters, detect SERP features, and check whether AI assistants cite your domain (these last ones use the paid SearchApi service). The README sets out three cost tiers: rule-based checks are free, a Haiku meta-description suggestion is about $0.0001 per URL, and a full Sonnet extraction and rewrite is about $0.006 per URL, so about $0.12 for 20 URLs.

What was reported (positive)

The author reports running it on their own sites and finding a title cannibalising the homepage, a query at average position 9.5 with no clicks, two missing internal links and an orphan page. No traffic or ranking change after fixing them is reported. During testing one AI assistant check returned an error and a timeout.

Limitations

Results are the author's own and cover only their sites; no before-and-after traffic is shown. The core audit runs in a visible browser and is not meant for servers or CI. Backlink scores are Claude's judgement of page content, not an industry metric, and the AI visibility check reflects model training cutoffs, so new sites often score zero. Needs Python and an Anthropic API key; the SERP and AI visibility modules need a paid SearchApi key, and the README links to SearchApi with tracking parameters as a developer ambassador. A premium folder mentions optional paid features (PDF reports, email delivery) without prices. The repo shows no release date; it was created in March 2026 and last updated in September 2026.

What you need

Python 3.11 or later, Playwright with Chromium, an Anthropic API key, optionally a SearchApi key, and a Search Console export for the extra modules.

Install the requirements, run 'playwright install chromium', set ANTHROPIC_API_KEY, put 10 to 20 of your most important URLs in input.csv and run the core audit interactively first so you can see where it stops.

Sources

Source published: unknown. Last reviewed here: October 27, 2026. Spot a mistake? Tell us.

Tools in this workflow

Get new workflows like this by email

A short selection of the most useful new workflows, including the ones that did not work. Free, and you can unsubscribe at any time.

Your email address

Related workflows