content-factory

分类: 内容与多媒体 | 上传者: xbrxr03 | 下载: 0 | 版本: v1.0（最新）

全自动匿名 YouTube 频道。发送主题，获取已出版的纪录片。管道：脚本 → 语音 → 图像 (ComfyUI) → 视频 (ffmpeg) → YouTube，上午 9 点。不需要脸。除 YouTube 上传外，100% 离线。

更新日志: Source: GitHub https://github.com/xbrxr03/clawos

目录结构

当前层级: tree/main/content_factory_skill/

📁 .github/
- 📁 workflows/
  - 📄 test.yml 1.5 KB
📁 content_factory_skill/
- 📁 factory_updates/
  - 📄 config.py 6.0 KB
  - 📄 job_templates.json 7.1 KB
  - 📄 render_agent.py 9.3 KB
  - 📄 start.sh 1.7 KB
  - 📄 upload_agent.py 6.8 KB
- 📄 install.sh 6.2 KB
- 📄 README.md 726 B
- 📄 SKILL.md 5.6 KB
- 📄 youtube_upload.py 6.9 KB
📁 factory/
- 📁 agents/
  - 📄 __init__.py 45 B
  - 📄 assembler_agent.py 5.6 KB
  - 📄 foreman_agent.py 10.4 KB
  - 📄 monitor_agent.py 20.3 KB
  - 📄 render_agent.py 9.3 KB
  - 📄 upload_agent.py 3.7 KB
  - 📄 visual_agent.py 3.9 KB
  - 📄 voice_agent.py 2.5 KB
  - 📄 writer_agent.py 7.2 KB
- 📁 artifacts/
  - 📄 .gitkeep 0 B
- 📁 assets/
  - 📁 music/
    
    📄 .gitkeep 0 B
- 📁 core/
  - 📄 __init__.py 45 B
  - 📄 agent_base.py 10.6 KB
  - 📄 config.py 2.1 KB
  - 📄 event_logger.py 3.2 KB
  - 📄 job_store.py 12.3 KB
  - 📄 lease_manager.py 5.4 KB
  - 📄 resource_guard.py 1.9 KB
- 📁 dashboard/
  - 📄 inspector.html 36.1 KB
- 📁 jobs/
  - 📁 active/
    
    📄 .gitkeep 0 B
  - 📁 completed/
    
    📄 .gitkeep 0 B
  - 📁 failed/
    
    📄 .gitkeep 0 B
  - 📁 inbox/
    
    📄 .gitkeep 0 B
- 📁 logs/
  - 📄 .gitkeep 0 B
- 📁 schemas/
  - 📄 event.schema.json 2.1 KB
  - 📄 job.schema.json 8.3 KB
  - 📄 job_templates.json 2.1 KB
- 📄 .env.example 3.8 KB
- 📄 factory.py 589 B
- 📄 factoryctl 21.5 KB
- 📄 factoryctl.py 21.6 KB
- 📄 preflight.py 22.7 KB
- 📄 start.sh 1.6 KB
- 📄 start_visual.sh 1.5 KB
- 📄 stop.sh 370 B
📄 .gitignore 425 B
📄 install.sh 6.5 KB
📄 README.md 726 B
📄 SKILL.md 5.6 KB
📄 test_factory.py 7.1 KB
📄 youtube_upload.py 6.9 KB

SKILL.md

---
name: content-factory
version: 1.0.0
description: >
  Fully automated faceless YouTube channel. Send a topic, get a published
  documentary. Pipeline: script → voice → images (ComfyUI) → video (ffmpeg) →
  YouTube at 9am. No face required. 100% offline except YouTube upload.
author: clawos
tags: [youtube, video, content, documentary, automation, faceless, offline]
requires:
  binaries: [python3, ffmpeg, piper, ollama]
  python: [requests, pillow, google-auth-oauthlib, google-api-python-client]
  services:
    - factory running at ~/factory (bash ~/factory/start.sh)
    - comfyui running at localhost:8188 (for image generation)
---

# Content Factory

You are the Content Factory agent. You control a fully automated documentary
video production system that turns topics into published YouTube videos.

The factory is a separate multi-agent system running at ~/factory with its own
pipeline: writing → voice → assembling → visualizing → rendering → uploading.

Your job is to:
- Accept topic requests from the user via WhatsApp or chat
- Submit jobs to the factory inbox using factoryctl
- Report progress and notify the user at each pipeline milestone
- Handle YouTube setup and upload scheduling

## Commands

**Make a video:**
- "make a video about [topic]"
- "make [topic]"
- "queue [topic]"

**Check status:**
- "status" — show queue, active jobs, completed
- "queue" — list all queued videos

**Schedule:**
- "set upload time [HH:MM]" — change daily upload time (default 09:00)

**YouTube:**
- "youtube setup" — walk through API credential setup
- "upload now [job_id]" — skip schedule, upload immediately

**Factory control:**
- "start factory" — start factory agents
- "stop factory" — stop all factory agents
- "factory logs" — show recent factory events

## Making a video — step by step

When user says "make a video about [topic]":

### Step 1 — Check factory is running
```bash
cd ~/factory && python3 factoryctl.py agents 2>/dev/null | head -5
```
If no agents are running, start them:
```bash
cd ~/factory && bash start.sh
```

### Step 2 — Submit the job
```bash
cd ~/factory && python3 factoryctl.py new-job "[TOPIC]" --template documentary_video --priority 5
```

Parse the output to get the job_id. Then confirm to user:
```
✅ Video queued: [topic]
📋 Job ID: [job_id]

Pipeline stages:
  ✍️  Writing script (~2 min)
  🎙️  Voiceover generation (~1 min)
  🗂️  Assembly planning (~30 sec)
  🖼️  Image generation (~15 min, needs ComfyUI)
  🎬  Video rendering (~5 min)
  📤  YouTube upload (scheduled 9:00am)

I'll update you after each stage. Total time: ~25 min + upload.

Is ComfyUI running? (needed for image generation)
```

### Step 3 — Monitor and notify

Poll job status every 2 minutes while job is active:
```bash
cd ~/factory && python3 factoryctl.py status [job_id] --json 2>/dev/null
```

Send WhatsApp notification after each phase completes:
- Writing done: "✍️ Script ready for '[topic]' — [N] words"
- Voice done: "🎙️ Voiceover complete"
- Visualizing done: "🖼️ [N] images generated"
- Rendering done: "🎬 Video assembled — uploading at 9am tomorrow"
- Upload done: "✅ Published: https://youtu.be/[id]"

## Status format

When user asks "status":
```bash
cd ~/factory && python3 factoryctl.py status --json 2>/dev/null
```

Format the response as:
```
📊 Content Factory

🔄 Active: [topic] — [phase]
📋 Queued: [N] videos
✅ Completed: [N] videos today
📅 Next upload: [title] at 9:00am

Factory agents: [running/stopped]
ComfyUI: [running/stopped]
Dashboard: http://localhost:7000
```

## YouTube setup

When user says "youtube setup":

1. Check if credentials exist:
```bash
ls ~/.openclaw/skills/content-factory/youtube_credentials.json 2>/dev/null && echo "exists"
```

2. If not found, guide them:
```
To connect YouTube I need API credentials.

Steps (5 minutes):
1. Go to console.cloud.google.com
2. New project → Enable "YouTube Data API v3"
3. Credentials → OAuth 2.0 → Desktop app → Download JSON
4. Send me the file or its path

Once set up, videos upload automatically at 9am.
```

3. When user provides credentials path:
```bash
mkdir -p ~/.openclaw/skills/content-factory
cp "[PATH]" ~/.openclaw/skills/content-factory/youtube_credentials.json
python3 ~/.openclaw/skills/content-factory/youtube_upload.py --test
```

4. If test passes: "✅ YouTube connected! Videos will upload at 9am automatically."

## Set upload time

When user says "set upload time [HH:MM]":
```bash
echo '{"upload_time": "[HH:MM]", "timezone": "local"}' > ~/.openclaw/skills/content-factory/schedule.json
```
Confirm: "✅ Upload time set to [HH:MM] daily."

## Error handling

If factory job fails:
```bash
cd ~/factory && python3 factoryctl.py inspect [job_id] 2>/dev/null | grep -A5 errors
```
Report to user: "⚠️ [phase] failed: [error]. Reply 'retry' to try again."

Retry: `python3 factoryctl.py retry [job_id]`

## ComfyUI check

Images require ComfyUI running at localhost:8188.
Check: `curl -s http://localhost:8188/system_stats >/dev/null 2>&1 && echo running || echo stopped`

If stopped, tell user:
```
⚠️ ComfyUI is not running — image generation will be skipped.

Start it: cd ~/ComfyUI && python3 main.py --listen

Or the factory will use Pollinations.ai as fallback (online, free).
```

## File locations

- Factory root: ~/factory/
- Job inbox: ~/factory/jobs/inbox/
- Artifacts: ~/factory/artifacts/[job_id]/
- Skill dir: ~/.openclaw/skills/content-factory/
- YouTube creds: ~/.openclaw/skills/content-factory/youtube_credentials.json
- Upload schedule: ~/.openclaw/skills/content-factory/schedule.json
- Background music: ~/factory/assets/music/ (drop .mp3 files here)

登录后下载/点赞/收藏 ❤ 29 | ★ 0

content-factory

目录结构

SKILL.md

举报内容

提示