Drives a real browser to open sites, fill forms, click, take screenshots and extract data.
vercel-labs/agent-browser
Gives AI agents a command-line browser so they can navigate, click, fill forms, scrape and test websites.
- 12 skills
- 50.7k stars
- Reached #1 on GitHub's daily trending
The 5 most installed
Ordered by installs on skills.sh- agent-browser1.6M installs
- derive-client120 installs
Records a site's network traffic and generates a standalone client or CLI that calls its API directly.
- modern-web-guidance61 installs
Applies modern web best practices to frontend code, then checks the result in a real browser.
- webmcp-gen53 installs
Builds and validates experimental WebMCP tools that expose a web page's workflows to agents.
- vercel4 installs
Opens Vercel preview and production deployments that sit behind login protection or feature flags.
The other 7 skills
Runs browser automation sessions on AWS Bedrock AgentCore cloud browsers using your AWS credentials.
Explains the basic browser workflow for navigating pages, clicking, filling forms, extracting data and troubleshooting.
Explores a web app to find bugs and UX issues, with screenshots and steps to reproduce each one.
Controls desktop apps like VS Code, Slack or Discord through browser automation to interact with or test them.
Checks unread channels, searches conversations and sends messages in Slack through browser automation.
Writes and edits developer docs such as getting-started guides, tutorials, API references and migration guides.
Runs headless Chrome inside Vercel Sandbox microVMs for browser automation from Vercel-deployed apps.