
A browser automation plugin for Claude Code that lets Claude control your browser to test and verify your work as you develop.
Key features:
/plugin marketplace add sawyerhood/dev-browser
/plugin install dev-browser@sawyerhood/dev-browser
Restart Claude Code after installation.
Copy the skill to your skills directory:
# For Amp: ~/.claude/skills | For Codex: ~/.codex/skills
SKILLS_DIR=~/.claude/skills # or ~/.codex/skills
mkdir -p $SKILLS_DIR
git clone https://github.com/sawyerhood/dev-browser /tmp/dev-browser-skill
cp -r /tmp/dev-browser-skill/skills/dev-browser $SKILLS_DIR/dev-browser
rm -rf /tmp/dev-browser-skill
Amp only: Start the server manually before use:
cd ~/.claude/skills/dev-browser && npm install && npm run start-server
The Chrome extension allows Dev Browser to control your existing Chrome browser instead of launching a separate Chromium instance. This gives you access to your logged-in sessions, bookmarks, and extensions.
Installation:
extension.zip from the latest release~/.dev-browser-extension)chrome://extensionsWhen active, the extension connects to Dev Browser's server and allows automation of your current Chrome tabs. Tell your agent to "use the extension" or "use my existing Chrome" to use it instead of the built-in browser.
To skip permission prompts, add to ~/.claude/settings.json:
{
"permissions": {
"allow": ["Skill(dev-browser:dev-browser)", "Bash(npx tsx:*)"]
}
}
Or run with claude --dangerously-skip-permissions (skips all prompts).
Just ask Claude to interact with your browser:
"Open localhost:3000 and verify the signup flow works"
"Go to the settings page and figure out why the save button isn't working"
| Method | Time | Cost | Turns | Success |
|---|---|---|---|---|
| Dev Browser | 3m 53s | $0.88 | 29 | 100% |
| Playwright MCP | 4m 31s | $1.45 | 51 | 100% |
| Playwright Skill | 8m 07s | $1.45 | 38 | 67% |
| Claude Chrome Extension | 12m 54s | $2.81 | 80 | 100% |
See dev-browser-eval for methodology.
| Approach | How It Works | Tradeoff |
|---|---|---|
| Playwright MCP | Observe-think-act loop with individual tool calls | Simple but slow; each action is a separate round-trip |
| Playwright Skill | Full scripts that run end-to-end | Fast but fragile; scripts start fresh every time |
| Dev Browser | Stateful server + agentic script execution | Best of both: persistent state with flexible execution |
MIT
$ claude mcp add dev-browser \
-- python -m otcore.mcp_server <graph>