playwright-manual-auth

Inspect authenticated web pages after manual login in an isolated headed browser.

Updated Aug 5, 2026
One-click install
npx skills add https://github.com/ahsanghalib/ai-workflow --skill playwright-manual-auth-ahsanghalib
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: playwright-manual-auth
Source: https://github.com/ahsanghalib/ai-workflow/tree/main/skills/playwright-manual-auth
Command: npx skills add https://github.com/ahsanghalib/ai-workflow --skill playwright-manual-auth-ahsanghalib

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Inspecting login-required websites with an AI assistant is risky: credentials, MFA codes, cookies, and session tokens can leak, and automated actions can accidentally change account state. This Skill provides a strict, read-only workflow where the user performs login manually in a fresh isolated browser while the agent never touches authentication material. ## Core Features & Use Cases - Manual Login Gate: The user completes login, MFA, and consent prompts directly in a headed isolated browser; the agent never sees or handles credentials. - Read-Only Inspection: Navigate approved origins and inspect rendered UI via snapshots and screenshots using only non-mutating controls like filters and pagination. - Hard Safety Boundaries: Prohibits credential handling, form submission, state-changing actions, arbitrary page code execution, and persistent browser attachment. - Use Case: You need to verify how a dashboard renders for a specific account. You approve the exact origin, log in yourself in the isolated browser, and the agent inspects the page and reports evidence-backed observations without touching your session data. ## Quick Start Ask the agent to inspect an approved page on your logged-in site after you complete the login manually in the isolated browser window.

Frequently Asked Questions about playwright-manual-auth

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I inspect a login-required website with an AI agent safely?▼

Approve the exact origins and a read-only goal, then complete the login yourself in a fresh headed isolated browser. The agent resumes only after you confirm login and inspects pages through snapshots and screenshots without ever touching credentials.

How does Playwright manual authentication work with an AI assistant?▼

The agent opens the approved sign-in page in a headed isolated browser and waits while you complete login, MFA, and consent prompts directly. It never inspects or interacts with those steps, then performs only non-mutating navigation and inspection afterward.

Can the agent enter my credentials or handle MFA codes?▼

No. The workflow strictly prohibits the agent from requesting, entering, reading, copying, retaining, or transmitting credentials, MFA data, cookies, tokens, or storage state. All authentication happens manually in the browser window by the user.

Does this work with any AI coding harness or only specific tools?▼

It is harness-agnostic and works with any environment that provides a configured headed isolated browser or Playwright adapter. It does not assume a specific CLI, provider tool namespace, or approval API.

What actions are blocked during authenticated browser inspection?▼

Form submissions, sending messages, creating or modifying content, purchases, settings changes, file uploads or downloads, arbitrary page code execution, and attaching to persistent browser profiles are all prohibited. Only non-mutating navigation and inspection are allowed.

When should I not use manual authenticated browser inspection?▼

Avoid it for public websites, credential entry automation, persistent sessions, durable regression testing, or any state-changing workflow. It is designed only for exploratory read-only inspection and does not provide regression coverage.