Capture All Pages

Documents

capture-all-pages

Capture All Pages is an Agent Skill for documents published in the majiayu000/claude-skill-registry repository on GitHub: a SKILL.md instruction file an AI agent can load to do this task.

Capture screenshots of all manual pages at high resolution for translation verification. Use when verifying that translations match page images, checking layout correctness, or inspecting for content mismatches.

View on GitHub ↗

QUICK START

How to use this skill

Bring this guide into your coding agent with a prompt tailored to the tool you use.

Open your project in Codex.
Copy and paste the prompt below.
Review files and risks before approving any action.

Prompt to paste

Inspect the Agent Skill at https://github.com/majiayu000/claude-skill-registry/blob/HEAD/skills/data/capture-all-pages/SKILL.md and all supporting files. Report provenance, permissions, scripts, dependencies, and security risks. Propose an exact installation into .agents/skills/capture-all-pages, but wait for my approval before writing files or running scripts.

Copying this prompt does not install or run the skill. Review third-party files before use.

Capture All Pages

Capture all pages of the manual viewer at high resolution (2000x1600) for visual verification of translations.

Purpose

This skill captures screenshots of all pages in the manual viewer to verify that:

Translations match the page images
Layout is correct
No content mismatches or missing translations

Quick Start

Prerequisites:

Dev server running on port 3100 (pnpm dev)
Playwright installed globally or in project

Usage:

# Capture all pages (default: 30 pages)
node .claude/skills/capture-all-pages/scripts/capture.js

# Capture specific number of pages
node .claude/skills/capture-all-pages/scripts/capture.js --pages 280

# Custom base URL
node .claude/skills/capture-all-pages/scripts/capture.js --base-url http://zmanuals.localhost:3100/manuals/oxi-one-mk2/page

Configuration

The capture script supports these options:

--pages <number> - Total pages to capture (default: 30)
--base-url <url> - Base URL for pages (default: http://zmanuals.localhost:3100/manuals/oxi-one-mk2/page)
--output-dir <path> - Custom output directory (default: __inbox/captures-{timestamp})

Default settings:

Viewport: 2000x1600 (high resolution for detail)
Output: __inbox/captures-{YYYYMMDD}-{HHMMSS}/
Format: PNG files named page-001.png to page-NNN.png
Timeout: 15 seconds per page
Wait after load: 2 seconds

Execution Workflow

Verify dev server is running

curl -s -o /dev/null -w "%{http_code}" http://zmanuals.localhost:3100/manuals/oxi-one-mk2/page/1
# Should return 200

Run capture script
```
node .claude/skills/capture-all-pages/scripts/capture.js
```
The script will:
- Create timestamped output directory
- Launch headless browser
- Capture each page sequentially
- Report progress and errors
- Save summary.json with results
Check results
```
ls -lh __inbox/captures-*/page-*.png
```

Output Structure

__inbox/captures-20260103-163245/
├── page-001.png (2000x1600)
├── page-002.png (2000x1600)
├── page-003.png (2000x1600)
...
├── page-030.png (2000x1600)
└── summary.json (capture results)

summary.json contains:

Timestamp and configuration
Success/failure counts
Error details for failed captures

Notes

All screenshots saved to project's __inbox/ directory (gitignored)
Session timestamp prevents overwriting previous captures
Script uses headless Chromium (no visible browser window)
Failed page captures are logged but don't stop the script
High resolution (2000x1600) allows detailed inspection of translations