Crawlbase MCP lets Codex scrape any URL and return its raw HTML, clean Markdown, or a screenshot. Powered by Crawlbase's infrastructure with JS rendering, automatic proxy rotation, and built-in anti-bot bypassing.
Install
npx skills add crawlbase/crawlbase-codex-plugingh skill install crawlbase/crawlbase-codex-plugingit clone --depth 1 https://github.com/crawlbase/crawlbase-codex-pluginSkills in this collection
- Crawl HtmlCrawlbaseCrawl a URL and return its full raw HTML content. Supports JS rendering, custom user agents, device emulation, geo-targeting, and optional screenshot capture.
- Crawl MarkdownCrawlbaseCrawl a URL and extract clean, readable Markdown content. Strips ads, navigation, and boilerplate — returns only the main content.
- Crawl ScreenshotCrawlbaseTake a full-page or viewport screenshot of any URL. Supports desktop and mobile viewports, custom dimensions, and configurable page-load wait times.
- Storage Bulk DeleteCrawlbaseDelete up to 100 stored pages from Crawlbase Cloud Storage in one call. Irreversible.
- Storage Bulk GetCrawlbaseRetrieve up to 100 stored pages from Crawlbase Cloud Storage in one call, optionally auto-deleting each item after retrieval.
- Storage CountCrawlbaseReturn the total number of documents in Crawlbase Cloud Storage for a token.
- Storage DeleteCrawlbaseDelete a single stored page from Crawlbase Cloud Storage by rid.
- Storage GetCrawlbaseRetrieve a single previously-stored page from Crawlbase Cloud Storage by rid or url. Returns the stored body as JSON, HTML, or Markdown.
- Storage ListCrawlbaseList rids currently in Crawlbase Cloud Storage, with scroll-based pagination (up to 1000 rids per call). Useful for enumerating everything stored under a token.