datum-cloud/Site Scrub

AI-assisted SEO & meta review for static sites. Scans built HTML with Cheerio, sends a compact report to an AI model (Claude, GPT, or Gemini), posts a sticky PR comment on pull requests and opens a weekly audit issue on scheduled/manual runs.

View on GitHub

Trust Signals

Scorecard Score
not yet scored
Maintenance Recency
Stale
License
None
namedescriptionrequireddefault
providerAI provider: 'anthropic' (default), 'openai', or 'gemini'.noanthropic
api-keyAPI key for the selected provider.no""
anthropic-api-keyDeprecated — use `api-key` (kept for backward compatibility with provider: anthropic).no""
base-urlOptional API base URL override (e.g. an OpenAI-compatible gateway like OpenRouter or Groq together with provider: openai).no""
site-urlProduction origin of the site (e.g. https://www.example.com). Used to recognise internal links and render absolute links in the report.yes
dist-dirDirectory containing the built HTML output.nodist
pages-dirSource directory whose files map 1:1 to routes (used for changed-file detection and broken-link validation).nosrc/pages
content-dirContent-collection directory whose files map to routes (used for changed-file detection).nosrc/content
scan-mode'auto' (changed-only on pull_request, full otherwise), 'changed-only', or 'full'.noauto
config-fileOptional path to a JSON config with excludePaths / excludeFiles / excludeFilePatterns.no""
broad-change-patternOptional regex; when a changed file matches it, the PR scan is skipped as a "broad change". Defaults to an Astro-oriented pattern (layouts, components, styles, config).no""
modelModel id override (defaults to the model pinned in the action for the selected provider).no""
max-pagesCap on pages analyzed (0 = unlimited).no0
github-tokenToken used to post the PR comment and manage the weekly audit issue.no${{ github.token }}

no outputs