agent-browser
Fast browser automation CLI for agents
About
A thin stub that points the agent at Vercel's agent-browser CLI, which drives Chrome through accessibility snapshots to open pages, fill forms, click, screenshot and run exploratory QA. The full workflow is loaded at runtime with agent-browser skills get core.
Install
Works with Claude Code, Codex, Cursor, Gemini CLI, GitHub Copilot, Goose, OpenCode, Windsurf
Trust score
50Scored from what the skill's own files let it do. How it works
- No bundled scripts
- Runs shell commands
- Doesn’t write files
- Uses the network
- Asks for credentials
Requires a global npm install plus a Chrome download, and its real instructions come from the installed CLI at runtime rather than this folder. It browses arbitrary sites and covers logging in and Slack automation.
Reviewed from source on Oct 15, 2026. Read the files yourself before installing.