Headless browser automation via the agent-browser CLI — snapshot + @ref workflow, ~200-400 tokens/interaction vs Playwright MCP. Use when automating a browser, scraping a JS-rendered page, filling forms, testing a web app, logging into a site, capturing screenshots, or introspecting a React app. Triggers on agent-browser, browser automation, scrape, web scraping, fill form, click button, screenshot, E2E test, snapshot refs, react devtools, web vitals
$npx -y skills add ariegoldkin/claude-forge --skill agent-browserInstalls into the current project.
Run `npx skills use "https://github.com/ariegoldkin/claude-forge" --skill "ariegoldkin/claude-forge/agent-browser"` and follow the generated skill instructions now. Read its complete output, redirecting it to a temporary file first if necessary. Resolve relative paths from the supporting-files directory it provides.
Use the skills in "https://github.com/ariegoldkin/claude-forge" that are relevant to the current task. Run `npx skills add "https://github.com/ariegoldkin/claude-forge"` and select the relevant skills, then follow their instructions.