- Home
- Skills
- Testing & Debugging
- Local Web Interface Probe
Works with the AI Agents you already use
Local Web Interface Probe
Drive, inspect, and debug local web apps with Playwright to capture screenshots, traces, and runtime evidence.
Free
Local Web Interface Probe
Example session with this skill installed
My Vite app is at ~/projects/taskflow and npm run dev serves on port 5173. The 'Create Project' button sometimes does nothing after the modal opens. Start the server if it's not up, open the app headless, go through the modal flow with a test project name, and tell me what actually happens. Capture console messages, failed requests, and full-page screenshots before and after the click, and if the save fires an API call, log its response. Leave the probe script in the project root so I can rerun it, and make sure nothing is left running when you're done.
- Read your context and instructions
- Compiled the local web interface
Summary
Probed the Create Project flow on http://127.0.0.1:5173 with a headless Chromium session. The button fails silently whenever the backend on :8000 is down — the modal's submit handler awaits POST /api/projects, the request fails, and no error branch renders.
| Step | Expected | Observed |
|---|---|---|
| Open app | Title + modal trigger render | OK, 0 console errors |
| Click 'Create Project' | Modal becomes visible | OK after ~120 ms |
| Fill name, submit | POST /api/projects returns 200 | ECONNREFUSED, unhandled promise rejection |
| Success feedback | Toast or redirect | Nothing; button stuck disabled |
| Evidence | Before/after captures | /tmp/before-click.png, /tmp/after-click.png |
Also saved: /tmp/probe/trace.zip and console.log. Reusable probe script written to ~/projects/taskflow/probe_create_project.py — rerun with python probe_create_project.py; it boots both servers if needed and shuts them down on exit.
Next steps
- Add an error branch to the modal submit handler so a failed POST re-enables the button and shows a toast
- Boot the API alongside the frontend in the dev script so the UI is never orphaned
- Re-run the probe script after the fix as a regression smoke check
Connects securely to your tools. The creator never sees your data.
About this skill
The problem
Static code analysis and raw HTML inspection fail to capture the behavior of modern SPAs and hydrated interfaces. You cannot debug what you cannot see, especially when UI elements only appear after JavaScript execution, routing, or specific user interactions.
What it does
- Launches headless Playwright instances to drive and inspect locally running web applications.
- Manages local server lifecycles, ensuring dev servers start and stop alongside the browser session.
- Captures runtime evidence including screenshots, full-page traces, console logs, and network failures.
- Discovers interactive elements like buttons, inputs, and links by evaluating the live, rendered DOM.
- Handles complex interactions such as dialogs, popups, and authenticated states through storage state reuse.
Frameworks & tools
Playwright, Python, Node.js dev servers, Chromium.
Why this beats prompting it yourself
General LLMs often suggest brittle CSS selectors or unreliable network-idle waits that cause flaky tests. This skill provides resilient interaction patterns, specific lifecycle management for local processes, and robust inspection scripts that extract the actual rendered state instead of guessing from source code.
Use cases
- Verify UI changes in a local React or Vue dev server before committing code.
- Debug a broken form submission by capturing the full network and console trace.
- Perform smoke tests on a local site across desktop and mobile viewports.
- Automate multi-step setups in a local dashboard to bypass repetitive manual navigation.
Known limitations
Requires local installation of Playwright and browser binaries. Not designed for complex CI/CD test suite architecture.
Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Testing & Debugging.
Requires Playwright and browser binaries installed locally. Best for one-off probes and debugging rather than building complete test frameworks.
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 buying a skill.
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
4 uses
Times used to date
Free to use
No account required to browse
Trust & safety
Security scan
Security scan passed 2 months ago
- Free with an Agensi account