Connect Playwright MCP to Gemini CLI
Browser automation over MCP from the Playwright team. Agents use it to open the built UI, take screenshots and check layouts against the design.
MCP serverOfficialChecked 2026-11-16
Before you start
- Gemini CLI installed (see its MCP documentation (opens in a new tab)).
- Playwright MCP is available on macOS, Windows and Linux.
Step 1: Add Playwright MCP to Gemini CLI
Add it from the terminal
Run this in your project folder. It saves the server to the project's
.gemini/settings.json.Terminal gemini mcp add playwright npx -- -y @playwright/mcp@latestOr edit the config file
Add this to
.gemini/settings.jsonin your project, or to~/.gemini/settings.jsonto use it in every project..gemini/settings.json { "mcpServers": { "playwright": { "command": "npx", "args": [ "-y", "@playwright/mcp@latest" ] } } }Step 2: Check it's connected
Terminal gemini mcp listIn a Gemini CLI session, run
/mcpto see each server, its status and its tools.Step 3: Try a first request
Ask Gemini CLI something that only works with Playwright MCP connected:
Say to Gemini CLI Open http://localhost:3000, take a screenshot at 375 px wide and tell me what looks broken.
If it doesn't work
- Since June 2026, Gemini CLI works with a paid Gemini API key or an enterprise licence. If you signed in with a personal Google account, use Google Antigravity instead.
gemini mcp addsaves to the project's.gemini/settings.jsonby default. Add--scope userto save it to~/.gemini/settings.jsonfor every project.- Local servers only show as connected in folders you trust. Run
gemini trustin the project folder. - Gemini CLI hides environment variables that look like secrets from servers unless the server's
envblock lists them.






