agent-browser · brianlovin — Claude Code skill: browser automation CLI
A skill that hands your coding agent a browser CLI for opening pages, filling forms, clicking elements and taking screenshots.
npx skills add brianlovin/agent-config --skill agent-browser -a claude-codeWhat agent-browser does
The skill exposes a set of agent-browser commands that drive Chrome or Chromium over CDP. Typical loop: open a URL, run agent-browser snapshot -i to get element refs such as @e1, then act on those refs. Re-snapshot after any navigation or DOM change, since old refs stop matching.
Calls share one browser through a background daemon, so chained commands with && are fine — open, wait for network idle, grab a screenshot in one shell line. Keep snapshot separate only when you need its output first. Auth is covered too: persistent profiles, named sessions, importing cookies from your running Chrome, an encrypted vault for credentials.
Before anything works you need the CLI installed (npm i -g agent-browser, brew install agent-browser or cargo install agent-browser) and Chrome pulled down with agent-browser install. Saved state files hold tokens in plaintext, so add them to .gitignore.
Screenshots come in plain, full-page and annotated flavours, where elements get numbered labels. There is also device emulation for mobile viewports.
What it helps with
- submit a web form without touching a mouse
- capture a full-page screenshot for a report
- log into an app once and reuse the session later
- pull element text off a page using snapshot refs
- download a file that only appears after a button click
Example prompts for the agent
- Open https://example.com/form, fill in the fields and submit it
- Take a full-page screenshot of the dashboard into
page.png - Log in using
--state ./auth.json, then confirm the profile page loads
How to install agent-browser
1Skills CLI (Vercel)
npx skills add brianlovin/agent-config --skill agent-browser -a claude-code2Manually: put the files in place
~/.claude/skills/agent-browser/SKILL.mdCopy the whole skill folder; for one project use .claude/skills/agent-browser/
1Skills CLI (Vercel)
npx skills add brianlovin/agent-config --skill agent-browser -a codex2Manually: put the files in place
.agents/skills/agent-browser/SKILL.mdFor all projects use ~/.agents/skills/
1Skills CLI (Vercel)
npx skills add brianlovin/agent-config --skill agent-browser -a cursor2Manually: put the files in place
.cursor/skills/agent-browser/SKILL.mdCursor also reads .agents/skills/ and .claude/skills/
1GitHub CLI
gh skill install brianlovin/agent-config agent-browser2Manually: put the files in place
.github/skills/agent-browser/SKILL.md1Skills CLI (Vercel)
npx skills add brianlovin/agent-config --skill agent-browser -a gemini-cli2Platform CLI
gemini skills install https://github.com/brianlovin/agent-config --path skills/agent-browserManually: put the files in place
.devin/skills/agent-browser/SKILL.mdThe legacy .windsurf/skills/ path still works
Skills CLI (Vercel)
npx skills add brianlovin/agent-config --skill agent-browser -a opencodeClaude.ai
Zip the folder and upload it in Settings → Capabilities → Skills
Step by step, with commands for each platform: How to install skills in Claude Code: five ways
Questions about agent-browser
How do I install agent-browser in Claude Code?
Run npx skills add brianlovin/agent-config --skill agent-browser -a claude-code; the skill lands at ~/.claude/skills/agent-browser/SKILL.md.
What does agent-browser do?
It gives the agent browser commands for navigation, clicking, typing, waiting on page state, screenshots and downloads.
What else is required besides the skill?
The CLI itself plus a downloaded Chrome via agent-browser install. Without those the commands have nothing to drive.
Similar skills
All skills for Browser and scraping →agent-browser by other authors comes first.