{
  "markdown": "# Train_Prep — Claude Code Plugin\n\nA Claude Code plugin that encapsulates a complete **AI consulting training material preparation workflow** — from industry pain point research to delivering a Word report and interactive HTML presentation, all orchestrated with multi-agent parallelism.\n\n## What It Does\n\nGiven a client and industry context (+ optional reference PDFs), the `Train_Prep` skill will:\n\n1. **Research** (parallel agents) — Analyze industry pain points across 3 business phases, map them to AI solutions\n2. **Word Report** — Assemble a structured `.docx` solution proposal via pandoc\n3. **HTML Slides** — Generate an interactive, full-screen HTML presentation with keyboard/touch navigation\n\n## Demo Output\n\nThe skill was originally created to prepare training materials for Pearl River Delta intercity rail executives. It produced:\n- A 40-pain-point solution proposal Word report\n- A 34-slide interactive HTML training presentation\n\n## Installation\n\n### Option 1: Install from GitHub (recommended)\n\n```bash\n# Coming soon — marketplace listing\n```\n\n### Option 2: Manual Install\n\n```bash\n# Clone the repo\ngit clone https://github.com/leeon/claude-train-prep.git ~/.claude/plugins/local/train-prep\n\n# Register in ~/.claude/plugins/installed_plugins.json\n# Add to \"plugins\" object:\n# \"train-prep@local\": [{ \"scope\": \"user\", \"installPath\": \"~/.claude/plugins/local/train-prep\", \"version\": \"local\" }]\n\n# Enable in ~/.claude/settings.json\n# Add to \"enabledPlugins\": { \"train-prep@local\": true }\n```\n\n## Usage\n\nOnce installed, trigger the skill by saying:\n\n- `Train_Prep`\n- `给 [客户名] 准备 AI 赋能培训材料`\n- `Prepare training materials for [industry] client`\n- `制作培训方案，客户是 [X]，业务场景是 [Y]`\n\nClaude will ask for:\n1. Client/industry context\n2. Reference PDFs (your capability deck + case studies)\n3. Output directory\n\n## Prerequisites\n\n```bash\nbrew install poppler pandoc   # macOS\n# or\nsudo apt-get install poppler-utils pandoc   # Linux\n```\n\n## Workflow Architecture\n\n```\n┌─────────────────────────────────────────────────────┐\n│                    Train_Prep Workflow                │\n├─────────────────────────────────────────────────────┤\n│  Phase 1: Research (3 parallel agents)               │\n│    Agent A ──► Phase 1 pain points + solutions       │\n│    Agent B ──► Phase 2 pain points + solutions       │\n│    Agent C ──► Phase 3 pain points + impl. plan      │\n├─────────────────────────────────────────────────────┤\n│  Phase 2: Word Report                                │\n│    Assemble Markdown ──► pandoc ──► .docx            │\n├─────────────────────────────────────────────────────┤\n│  Phase 3: HTML Slides (4 parallel agents)            │\n│    Agent A ──► HTML framework + CSS + JS + cover     │\n│    Agent B ──► Phase 1 slides                        │\n│    Agent C ──► Phase 2 slides                        │\n│    Agent D ──► Phase 3 + impl. plan + closing        │\n│    Main ────► macOS-compatible assembly script       │\n└─────────────────────────────────────────────────────┘\n```\n\n## Plugin Structure\n\n```\nclaude-train-prep/\n├── .claude-plugin/\n│   └── plugin.json\n└── skills/\n    └── Train_Prep/\n        ├── SKILL.md                    # Core skill (triggers + workflow)\n        └── references/\n            └── workflow-guide.md       # Detailed code templates & checklist\n```\n\n## Key Technical Notes\n\n- **PDF extraction**: Uses `pdftotext` (more reliable than Claude's built-in PDF reader for text extraction)\n- **macOS compatibility**: `head -n -N` (negative line count) is not supported on macOS — the skill uses `lines=$(wc -l < file); head -n $((lines-N)) file` instead\n- **HTML assembly**: Multi-agent HTML parts use a strict interface contract (Part 1 leaves `slide-container` unclosed; Part 4 closes it with `</body></html>`)\n- **Dynamic slide counting**: Uses `MutationObserver` instead of hardcoded totals, enabling agents to work independently without coordination\n\n## License\n\nMIT License — see [LICENSE](LICENSE)\n\n## Contributing\n\nPRs welcome! Especially:\n- Additional industry examples in `references/`\n- Improved HTML slide templates\n- Better ROI calculation frameworks\n",
  "bytes": 4099,
  "sha": "2224ccc89d81e96630475d27c5d294ebf1c8623b6e3cba3531e8e359a4d516db",
  "repo_slug": "gzleeon/claude-train-prep",
  "fonte": "repo",
  "truncated": false,
  "api": "https://agentalog.com/api/listings/plg_gzleeon_claude_train_prep_claude_train_p_42e1d718/readme"
}