# Automate Web Checks with Playwright

Manual page checks slow down web testing and content review. This skill opens pages in Chromium and returns screenshots, text, or HTML.

## Install

```bash
npx skillstore add enactprotocol/enact-playwright
```

## Metadata

- Status: approved
- Slug: enactprotocol-enact-playwright
- Version: 1.0.0
- Author version: 1.0.0
- Skillstore revision: r1
- Version status: valid
- Tree hash: c1bb9788a3f9fac54eb0920c3f6b975573c32984bb845d683c37da98f9330065
- Author: EnactProtocol
- GitHub username: EnactProtocol
- License: MIT
- Repository: https://github.com/EnactProtocol/enact/tree/main/test-tools/playwright
- Ref: a06681402992ceae98ba04d54cfd4ab004862696
- Supported tools: Claude, Codex, Claude Code
- Audit status: complete
- Agent install advisory: confirmation\_required
- Manual install advisory: allowed
- Artifact signature: available
- Audit attestation: unavailable
- Human verification: not\_verified
- Risk factors: external\_commands, filesystem, network
- Quality score: 38
- Quality tier: warning
- Public page: https://skillstore.pages.dev/skills/enactprotocol-enact-playwright
- Manifest: https://skillstore.pages.dev/api/skills/enactprotocol-enact-playwright/manifest

## Capabilities

- Navigates to a supplied URL with headless Chromium.
- Captures full-page screenshots and returns base64 output.
- Extracts text from the full page or a CSS selector.
- Extracts inner HTML from the selected element.
- Uses playwright-core with the Microsoft Playwright container image.
- Applies a 30 second page navigation timeout.

## Use Cases

- Capture Regression Evidence: Take a full-page screenshot of a staging page for review before release.
- Extract Page Copy: Collect visible text from a public page or a selected element for analysis.
- Inspect Rendered Markup: Retrieve inner HTML from a page section to debug rendered output.

## Prompt Templates

### Get Page Text

```
Open the target web page and return the visible text from the body.
```

### Capture a Screenshot

```
Navigate to the supplied URL and capture a full-page screenshot for review.
```

### Read a Specific Element

```
Open the page and return the text from the main heading selector.
```

### Inspect Rendered HTML

```
Navigate to the page and return the inner HTML for the selected component container.
```

## Limitations

- It does not click buttons, fill forms, or run multi-step browser flows.
- It does not authenticate to sites or manage browser sessions.
- Screenshot output is returned as base64 text, not as a saved image file.
- Invalid selectors or blocked pages surface as Playwright errors.

## Best Practices

- Use public or approved test URLs unless confidential access is required.
- Prefer narrow CSS selectors when extracting text or HTML.
- Review screenshot and HTML output before sharing it outside your team.

## Anti Patterns

- Do not target internal network hosts without explicit authorization.
- Do not pass secrets, tokens, or credentials inside page URLs.
- Do not use this skill for complex browser workflows that require clicks or login state.

## Security Audit

- Audited at: 2026-07-06T15:22:37.35\+00:00
- Summary: The static backtick, filesystem, and documentation URL findings are false positives after context review. Semantic review found meaningful risk from user-controlled command interpolation in SKILL.md and unrestricted browser navigation to caller-supplied URLs.

## Stats

- Views: 181
- Downloads: 11
- Favorites: 0
- Popularity score: 0