Podcastfy Clawdbot Skill

podcastfy-clawdbot-skill

A Clawdbot skill that generates a podcast-style MP3 from one or more URLs using the open-source **Podcastfy** project.

It is designed to be used inside a Clawdbot deployment and provides a wrapper script that:

• Creates/uses a local Python venv
• Installs/updates podcastfy
• Uses Gemini for script generation (LLM)
• Uses Edge TTS for speech synthesis (no paid TTS API required)

Note: This repository intentionally does **not** include any generated audio/transcripts or any secrets.

---

Features

• Generate a single MP3 from a single article URL
• Generate one MP3 from multiple URLs (e.g., 3 articles → 1 combined episode)
• Compatible with Clawdbot scheduled jobs (cron)
• Safer defaults: .env, .venv, and output/ are ignored

---

Requirements

• Linux/macOS (tested in Linux)
• Python 3.10+ (tested with Python 3.12)
• ffmpeg
• A valid GEMINI_API_KEY (Gemini API)

Install ffmpeg (Ubuntu/Debian):

sudo apt-get update && sudo apt-get install -y ffmpeg

---

Setup

1) Create a `.env`

Create a `.env` file in the skill directory:

GEMINI_API_KEY=YOUR_KEY_HERE
PODCASTFY_LLM_MODEL=gemini-2.5-flash

Optional (Edge TTS voices):

PODCASTFY_EDGE_VOICE_Q=en-US-JennyNeural
PODCASTFY_EDGE_VOICE_A=en-US-EricNeural

2) Run

Single URL:

cd skills/podcastfy-clawdbot
./scripts/podcastfy_generate.py --url "https://example.com/article"

Multiple URLs:

cd skills/podcastfy-clawdbot
PODCASTFY_LLM_MODEL=gemini-2.5-flash \
  ./scripts/podcastfy_generate.py \
  --url "https://example.com/a" \
  --url "https://example.com/b" \
  --url "https://example.com/c"

---

Output

On success, the script prints the generated MP3 path.

Typical output locations:

• output/audio/*.mp3
• output/transcripts/*.txt

---

Notes / Troubleshooting

• If Gemini returns 429 with limit: 0, it usually means your project/key has no usable quota (billing/quota not enabled