Bowser

Playwright Bowser

Purpose

Automate browsers using playwright-cli — a token-efficient CLI for Playwright. Runs headless by default, supports parallel sessions via named sessions (-s=), and doesn't load tool schemas into context.

Requirements

This skill drives the external Playwright Agent CLI (playwright-cli). It is not bundled with agent-fleet — install it once before using this skill (the guided setup checks for it when bowser is selected).

• Node.js 20+
• playwright-cli — install globally:

npm install -g @playwright/cli@latest
playwright-cli --help                 # verify it responds

Or run ad-hoc without a global install via npx playwright-cli .

Browsers are downloaded automatically on first use; to pre-install explicitly:

playwright-cli install-browser               # default (chromium)
playwright-cli install-browser firefox       # a specific browser
playwright-cli install-browser --with-deps   # include system dependencies

Full install docs:

When to use this vs Chrome DevTools MCP

This skill (playwright-cli) and the chrome-devtools-mcp pi extension are complementary, not redundant — they differ by tool model, not just headless-vs-headful (both can do either):

• bowser / playwright-cli (this skill) — CLI-driven, headless-first, parallel named sessions, token-efficient (no tool schemas in context). It needs only Bash, so it works as a dispatched subagent under --no-extensions. Use it for automated, parallel, or background runtime-UI evidence and scraping.
• web-debugger / chrome-devtools-mcp — live chrome_devtools__* tools (DOM/console/network/performance traces) for interactive headful debugging with a human in the loop. Those tools come from an always-on extension, so they are reachable in the main session or a coms peer, not a --no-extensions subagent. Use it to understand a failure in a running dev app.

Key Details

• Headless by default — pass --headed to open to see the browser
• Parallel sessions — use -s= to run multiple independent browser instances
• Persistent profiles — cookies and storage state preserved between calls
• Token-efficient — CLI-based, no accessibility trees or tool schemas in context
• Vision mode (opt-in) — set PLAYWRIGHT_MCP_CAPS=vision to receive screenshots as image responses in context instead of just saving to disk

Sessions

Always use a named session. Derive a short, descriptive kebab-case name from the user's prompt. This gives each task a persistent browser profile (cookies, localStorage, history) that accumulates across calls.

# Derive session name from prompt context:
# "test the checkout flow on mystore.com" → -s=mystore-checkout
# "scrape pricing from competitor.com"    → -s=competitor-pricing
# "UI test the login page"               → -s=login-ui-test

playwright-cli -s=mystore-checkout open https://mystore.com --persistent
playwright-cli -s=mystore-checkout snapshot
playwright-cli -s=mystore-checkout click e12

Managing sessions:

playwright-cli list                                     # list all sessions
playwright-cli close-all                                # close all sessions
playwright-cli -s= close                          # close specific session
playwright-cli -s= delete-data                    # wipe session profile

Quick Reference

Core:       open [url], goto , click , fill  , type , snapshot, screenshot [ref], close
Navigate:   go-back, go-forward, reload
Keyboard:   press , keydown , keyup 
Mouse:      mousemove  , mousedown, mouseup, mousewheel  
Tabs:       tab-list, tab-new [url], tab-close [index], tab-select 
Save:       screenshot [ref], pdf, screenshot --filename=f
Storage:    state-save, state-load, cookie-*, localstorage-*, sessionstorage-*
Network:    route , route-list, unroute, network
DevTools:   console, run-code , tracing-start/stop, video-start/stop
Sessions:   -s= , list, close-all, kill-all
Config:     open --headed, open --browser=chrome, resize  

Workflow

1. Derive a session name from the user's prompt and open with --persistent to preserve cookies/state. Always set the viewport via env var at launch:

PLAYWRIGHT_MCP_VIEWPORT_SIZE=1440x900 playwright-cli -s= open  --persistent
# or headed:
PLAYWRIGHT_MCP_VIEWPORT_SIZE=1440x900 playwright-cli -s= open  --persistent --headed
# or with vision (screenshots returned as image responses in context):
PLAYWRIGHT_MCP_VIEWPORT_SIZE=1440x900 PLAYWRIGHT_MCP_CAPS=vision playwright-cli -s= open  --persistent

1. Get element references via snapshot:

playwright-cli snapshot

1. Interact using refs from snapshot:

playwright-cli click 
playwright-cli fill  "text"
playwright-cli type "text"
playwright-cli press Enter

1. Capture results:

playwright-cli screenshot
playwright-cli screenshot --filename=output.png

1. Always close the session when done. This is not optional — close the named session after finishing your task:

playwright-cli -s= close

Configuration

If a playwright-cli.json exists in the working directory, use it automatically. If the user provides a path to a config file, use --config path/to/config.json. Otherwise, skip configuration — the env var and CLI defaults are sufficient.

{
  "browser": {
    "browserName": "chromium",
    "launchOptions": { "headless": true },
    "contextOptions": { "viewport": { "width": 1440, "height": 900 } }
  },
  "outputDir": "./screenshots"
}

Full Help

Run playwright-cli --help or playwright-cli --help for detailed command usage.

See the Playwright Agent CLI docs for full documentation.

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: chankov
• Source: chankov/agent-fleet
• License: MIT

Install and usage instructions live in the source repository linked above.