Podcast Discovery & Generation

Automated end-to-end podcast production pipeline. Discovers trending topics from configurable sources, researches them deeply, generates fact-checked scripts with citations, and produces audio via ElevenLabs TTS.

Triggers

Use this skill when user asks to:

• "Generate a podcast"
• "Make a podcast episode"
• "Discover podcast topics"
• "Create an audio episode about X"
• "Find topics for podcast"
• "Research and script a podcast"
• "Produce a podcast episode"

Quick Start

1. Configure

cd ~/.openclaw/skills/podcast
cp config.example.yaml config.yaml
# Edit config.yaml: add sources, interests, voice, storage

2. Discover Topics

python3 scripts/discover.py --config config.yaml --limit 10

3. Run Pipeline

python3 scripts/pipeline.py --config config.yaml --topic "Your Topic" --mode manual

Configuration

Minimal config.yaml:

sources:
  - type: rss
    url: https://aeon.co/feed.rss
    name: Aeon
  - type: hackernews
    min_points: 200

interests:
  - AI/Tech
  - Science

voice:
  voice_id: "<your-voice-id>"

storage:
  type: local
  path: ./output

Storage options:

• type: s3 — Upload to S3 (requires bucket, region)
• type: local — Save to local directory

Pipeline Stages

1. Discovery — Fetch and rank topics from sources
2. Research — Web search framework (OpenClaw worker populates)
3. Script — Generate script with LLM, enforce [Source: URL] citations
4. Verify — Cross-check claims against research sources
5. Audio — Strip citations, call ElevenLabs TTS
6. Upload — Save to S3 or local storage

Each stage can run standalone or as full pipeline.

Usage Examples

Discover only:

python3 scripts/discover.py --config config.yaml --limit 5 --output topics.json

Full pipeline (auto mode):

python3 scripts/pipeline.py --config config.yaml --mode auto

Specific topic:

python3 scripts/pipeline.py --config config.yaml --topic "AI Reasoning" --mode manual

Resume from stage:

python3 scripts/pipeline.py --config config.yaml --resume-from audio

Source Types

Built-in:

• rss — Generic RSS/Atom feed (any URL)
• hackernews — HN API with point/comment filters
• nature — Nature journal (sections: news, research, biotech, medicine)

Add custom RSS:

sources:
  - type: rss
    url: https://yourfeed.com/rss
    name: Your Source
    category: Your Category

Output Files

output/
├── discovery-YYYY-MM-DD.json      # Ranked topics
├── research-YYYY-MM-DD-slug.json  # Research data
├── script-YYYY-MM-DD-slug.txt     # Script with citations
├── verification-YYYY-MM-DD.json   # Fact-check report
├── tts-ready-YYYY-MM-DD-slug.txt  # Clean text for TTS
├── episode-YYYY-MM-DD-slug.mp3    # Final audio
└── pipeline-state-YYYY-MM-DD.json # Pipeline state

Integration with OpenClaw

For discovery: Run directly (no tools needed)