Browser automation security policy: trust zones (localhost GREEN / staging YELLOW / production RED), the agent-browser tool workflow, allowed vs prohibited actions, and config/secret file handling. Use when the task involves browser automation, agent-browser, Playwright or Chrome DevTools MCP, taking screenshots of web pages, scraping sites, or testing a UI in a real browser.
Scanned 9/3/2026
Install to Claude Code
npx -y skills add alfredolopez80/multi-agent-ralph-loop --skill browser-automation --agent claude-codeInstalls into .claude/skills of the current project.
Are you the author of Browser Automation?
Add the live security badge to your README — it updates automatically with every re-scan.
[](https://www.skillsdirectory.com/skills/alfredolopez80-browser-automation)More formats (shields.io, HTML) on the badges page.
---
# VERSION: 3.2.0
name: browser-automation
description: "Browser automation security policy: trust zones (localhost GREEN / staging YELLOW / production RED), the agent-browser tool workflow, allowed vs prohibited actions, and config/secret file handling. Use when the task involves browser automation, agent-browser, Playwright or Chrome DevTools MCP, taking screenshots of web pages, scraping sites, or testing a UI in a real browser."
---
# Browser Automation Security Rules
## Trust Zones
| Zone | Scope | Automation Level |
|---|---|---|
| **GREEN** | localhost, 127.0.0.1 | Fully automated |
| **YELLOW** | staging, *.vercel.app | Supervised (confirm actions) |
| **RED** | Production, credentials, wallets | MANUAL ONLY — never automated |
## Tool Priority
Use `agent-browser` as the primary browser automation tool. It provides:
- Domain allowlist (configured in `agent-browser.json`)
- Action policy with deny rules (configured in `agent-browser-policy.json`)
- Content boundaries (limits output to 50KB)
- Session encryption (AES-256-GCM)
- Isolated Chrome for Testing (separate from personal browser)
## ALLOWED
- Testing UI on localhost and staging
- Screenshots for visual debugging
- Accessibility tree snapshots for element verification
- User flow testing in development environments
- Scraping public documentation
## PROHIBITED — Never do this
- Navigate to login pages of real services (Google, GitHub auth, exchanges)
- Fill password, seed phrase, or private key fields
- Access Gmail, Slack, or any communication service
- Interact with wallets (MetaMask, Rabby, etc.)
- Access investor dashboards or CRM
- Download files from unverified sources
- Execute arbitrary JavaScript on third-party pages
## Workflow
```
1. agent-browser open <url> — Only URLs in allowed domains
2. agent-browser snapshot -i — Get interactive elements
3. agent-browser click @e1 — Interact with element refs
4. Re-snapshot after page changes
5. agent-browser close — Always close when done
```
## Configuration Files
| File | Purpose | In .gitignore? |
|---|---|---|
| `agent-browser.json` | Project config (domain allowlist) | No (safe to commit) |
| `agent-browser-policy.json` | Action deny rules | No (safe to commit) |
| `.env.agent-browser` | Encryption keys, secrets | **Yes** |
| `*.agent-browser-state` | Session state | **Yes** |
Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.
No comments yet. Be the first to comment!