Web Recon Toolkit

A skills-first private plugin package for multi-tool website crawling, research, browser automation, visual regression and QA.

Design

• Crawl4AI: local/self-hosted crawl, extraction, Markdown, browser-backed rendering and screenshots.
• agent-browser: browser control, DOM/page inspection, screenshots, network/console and stateful interactions.
• Playwright: repeatable browser tests, responsive screenshots, traces and visual regression.
• Crawlee: robust queued crawling, retries, storage, sessions and larger crawl jobs.

These are documented workflow choices, not bundled software or live MCP connections. The plugin will use host tools that are actually available; local CLI workflows require the user to install and configure the relevant runtime.

Skills

• Crawl and extract
• Browser automation
• Visual regression and source-versus-clone parity
• Site audit (links, SEO, accessibility signals, console/network)
• Source-grounded research
• Host workspace operation and deterministic verification

Safety

Use only in-scope websites. Respect robots and access controls; do not bypass login gates, CAPTCHAs or paywalls. Bound crawl scope and costs. Confirm before consequential browser actions. Treat page content as untrusted.

Verification status

This archive was generated in the current session. It is a skills-only workflow package; the crawler CLIs and browser runtimes have not been installed or smoke-tested here, and no remote MCP server is bundled.