- Home
- Skills
- Testing & Debugging
- Terminal Browser Operator
Works with the AI Agents you already use
Terminal Browser Operator
Drive a real Playwright browser from the command line — navigate, fill forms, extract data, capture screenshots and traces, and debug UI flows without...
Free
Terminal Browser Operator
Example session with this skill installed
I need to verify our signup funnel after yesterday's release. From the terminal, open https://app.acme-saas.com/signup in a headed browser, fill the email, password, and country fields using fresh accessibility refs, tick the terms checkbox, and submit. Confirm we land on /welcome and the success banner is actually visible — don't guess at selectors: snapshot before each action and re-snapshot after the redirect. Save a full-page screenshot and a PDF of the confirmation page under output/browser, and start a trace right before the submit click so I can replay the flow later. If a ref goes stale mid-run, recover gracefully and note it in your report.
- Read your context and instructions
- Compiled the terminal browser operator
- Generated the document
Summary
Drove the Acme signup funnel end to end from the terminal with snapshot-anchored Playwright CLI commands: filled the form, submitted, verified the redirect to /welcome, and captured screenshot, PDF, and trace artifacts for review.
| Step | Command | Result |
|---|---|---|
| 1 | browser open https://app.acme.dev/signup --headed | Live session, page loaded |
| 2 | browser snapshot | Refs mapped: email=e7, password=e8, country=e18, terms=e22, submit=e12 |
| 3 | browser fill e7 qa@acme.dev / fill e8 ******* | Both fields accepted |
| 4 | browser select e18 "United States" / check e22 | Dropdown and checkbox confirmed via re-snapshot |
| 5 | tracing-start → click e12 → tracing-stop | Redirect to /welcome, success banner e3 visible |
| 6 | browser screenshot + pdf | output/browser/screens/signup-welcome.png, welcome-report.pdf |
Next steps
- Open the trace in the Playwright trace viewer to inspect the 1.2s lag between submit and redirect
- Re-run once on a fresh build to confirm refs e7/e8 stay stable across the redesigned form
- Add the screenshot check to the release checklist before Friday's deploy
All refs came from live snapshots; re-snapshot before replaying on a different build.
signup-welcome.png
PDF · document
Example file from a real run - the skill writes it into your workspace.
Connects securely to your tools. The creator never sees your data.
About this skill
The problem
Controlling a web browser through an AI agent often leads to fragile scripts, stale element selectors, and a lack of visual feedback. Manual browser automation is slow, and writing full test suites for simple data extraction or form filling is overkill.
What it does
- Controls a real Playwright-powered browser directly from the terminal.
- Uses a snapshot-first workflow to discover stable element references before interacting.
- Captures visual evidence including screenshots, PDFs, and full execution traces.
- Manages complex UI states, multiple tabs, and form interactions through a CLI wrapper.
- Recovers from stale element references by re-mapping the DOM on demand.
Frameworks & tools
Node.js, Playwright, shell scripting, and npx.
Why this beats prompting it yourself
This skill implements a robust snapshot-action loop that prevents the agent from guessing selectors. It handles the boilerplate of browser context management and provides a structured way to recover when the UI changes, which raw prompting usually misses.
Use cases
- Automate repetitive form submissions or sign-up flows.
- Scrape data from dynamic, JavaScript-heavy websites that block simple requests.
- Generate PDF reports or screenshots of dashboards for automated reporting.
- Debug production UI flows by capturing traces and snapshots in a headless environment.
Known limitations
Requires Node.js and npm to be installed on the host system. The skill depends on a specific shell wrapper for optimal performance.
Requires Node.js/npx on the host and first-run network access to fetch the Playwright CLI. Exact command flags can shift between CLI versions, and it stays in CLI mode — it does not generate @playwright/test specs unless explicitly asked.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before any payment.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
2 uses
Times used to date
Free forever
No account required to browse
Trust & safety
Security scan
Security scan passed 1 month ago
- Free with an Agensi account
Requires
Permissions
File Scopes
Shell access is required to run the bundled browser CLI wrapper, the browser must reach user-specified URLs, and write access to an artifacts folder stores screenshots, PDFs, and traces.