postall

name	postall
description	Generate and publish social media posts via PostAll workflow. Supports multi-platform content generation (Twitter, LinkedIn, XHS) with AI-powered content creation and quality review.

PostAll - Multi-Platform Content Publishing

Generate and publish content to Twitter, LinkedIn, and Xiaohongshu via PostAll.

Quick Start

# 1. Clone and setup
git clone https://github.com/qingxuantang/postall.git
cd postall
docker-compose up -d

# 2. Configure your project
cp projects/example/project.yaml projects/myproject/project.yaml
# Edit project.yaml with your settings

# 3. Run content generation
docker exec postall python /app/run_topic_example.py

Workflow

1. Content Extraction (Optional)

For YouTube videos, extract content first:

# Download audio
yt-dlp -x --audio-format mp3 -o "/tmp/%(id)s.%(ext)s" "<youtube_url>"

# Transcribe with Whisper API
curl -s https://api.openai.com/v1/audio/transcriptions \
  -H "Authorization: Bearer $OPENAI_API_KEY" \
  -F file="@/tmp/<video_id>.mp3" \
  -F model="whisper-1" \
  -F response_format="text" > transcript.txt

2. Create Topic Script

Copy an existing script as template:

cp projects/example/run_topic_template.py projects/myproject/run_topic_<name>.py

Edit only these sections:

• TOPIC_NAME - Unique identifier (lowercase, no spaces)
• TOPIC - Your content/topic details
• RELATED_HITLIST - Related talking points (optional)

3. Run Pipeline

docker exec postall python /app/run_topic_<name>.py

The pipeline will:

1. Generate Twitter thread
2. Generate LinkedIn post
3. Generate article content
4. Run Director quality review
5. Generate cover image

4. Director Quality Gate

Check the Director review before publishing:

docker exec postall cat /app/output/single_topics/<topic>/director_review_report.md

Score	Decision	Action
≥7.0	Approve	Proceed to publish
<7.0	Reject/Revise	Review feedback and regenerate

5. Publish

# Copy clean_and_publish.py to container
docker cp clean_and_publish.py postall:/tmp/

# Twitter
docker exec postall python /tmp/clean_and_publish.py twitter <content_file> <image_file>

# LinkedIn
docker exec postall python /tmp/clean_and_publish.py linkedin <content_file> <image_file>

# Xiaohongshu
docker exec postall python /tmp/clean_and_publish.py xhs <cards_dir> <title>

Timeliness System

PostAll includes a timeliness context system to ensure generated content references current tools and trends.

# Update context
docker exec postall python -m postall.utils.timeliness_context refresh

# View current context
docker exec postall python -m postall.utils.timeliness_context show

Edit data/timeliness_manual_context.json to customize:

• current_hot_tools: Current trending AI tools
• current_trends: Current industry trends
• outdated_references: Tools no longer considered "cutting edge"

LinkedIn Token Management

LinkedIn tokens expire every 60 days. Use the auto-refresh script:

# Initial setup
export LINKEDIN_CLIENT_ID="your_client_id"
export LINKEDIN_CLIENT_SECRET="your_client_secret"
python3 linkedin_auto_refresh.py --init <access_token> <refresh_token>

# Test token
python3 linkedin_auto_refresh.py --test

# Publish with auto-refresh
python3 publish_linkedin.py <content_file> [image_file]

Critical Rules

1. NO parentheses - Can cause LinkedIn truncation issues
2. NO AI clichés - Avoid overused AI phrases
3. NO fabricated data - Only use real information from sources
4. Image prompts in English - Gemini cannot generate Chinese text
5. Twitter rate limits - If hit, reschedule for next day
6. No tweet numbering - Avoid "1/ 2/ 3/" format, looks AI-generated

Output Structure

output/single_topics/<topic>/
├── x-tweets/           # Twitter content
│   └── twitter_content.md
├── linkedin-posts/     # LinkedIn content
│   └── linkedin_content.md
├── wechat-posts/       # Article content
│   └── wechat_content.md
├── xhs-cards/          # Xiaohongshu cards (if generated)
│   ├── card_01.png
│   └── ...
└── image.png           # Cover image

Project Configuration

See projects/example/project.yaml for configuration options:

• Brand voice and style
• Platform-specific CTA links
• Content pillars
• Target audience

API Requirements

• Claude API (Anthropic) - Content generation
• Gemini API (Google) - Image generation
• OpenAI API (optional) - Whisper transcription
• Platform APIs - Twitter, LinkedIn, Xiaohongshu