seo-drift stores a snapshot of a page’s SEO-critical elements and diffs the live page against it. A baseline records the title, meta description, canonical, robots directives, H1 to H3 headings, JSON-LD schema, Open Graph tags, Core Web Vitals, HTTP status and hashes of the HTML and schema.
Comparison applies 17 rules at three severity levels, critical, warning and info, and points each finding to the skill that investigates it. Everything is kept in a local SQLite database.
What You Get
- A stored baseline with a snapshot ID and timestamp
- A JSON diff of triggered rules with old and new values, severity and recommended action
- A history of baselines and comparisons per URL
- An optional drift-report.html generated from a comparison
- Follow-up suggestions such as running the schema, technical or content skill
Common Use Cases
- Check that a deployment did not remove canonicals or add noindex
- Monitor key landing pages for title or schema edits
- Find out when a page changed during a traffic drop
- Catch Core Web Vitals regressions after a release
What Makes It Unique
It is a regression monitor, not a full audit: it normalizes URLs, keeps data local, and ties each severity to a response time. The fetch pipeline has SSRF protection, queries use parameters, TLS is always verified and there are no raw curl calls.
Under The Hood
The skill runs shell commands that start Python scripts through the claude-seo launcher. They fetch the page over the network, parse its HTML, optionally call the PageSpeed API for Core Web Vitals, hash the content and read and write a SQLite file at ~/.cache/claude-seo/drift/baselines.db, creating it on first use. The report command writes an HTML file. Requests to private addresses are refused. Fetched content is treated as untrusted data, and the frontmatter does not declare the shell, network or file access.