config.example.yml

# podwash configuration
# Copy to config.yml and customize

# Base URL where this service is accessible (used in RSS feed URLs)
base_url: "http://localhost:8080"

# Directory for SQLite database and audio files
data_dir: "./data"

# Feeds to monitor
feeds:
  - name: "My Favorite Podcast"
    url: "https://feeds.example.com/podcast.xml"
    slug: "my-favorite-podcast"
    poll_interval_minutes: 60

  # - name: "Another Podcast"
  #   url: "https://feeds.example.com/another.xml"
  #   slug: "another-podcast"
  #   poll_interval_minutes: 120

  # Optional: keep only source episodes whose title contains any of
  # these case-insensitive substrings. Useful when a podcast host
  # publishes multiple shows into one shared feed.
  #   title_includes:
  #     - "Midnight Boys"

# Processing settings
processing:
  # Whisper model: tiny, base, small, medium, large-v3
  whisper_model: "base"
  # Whisper compute type: int8, float16, float32
  whisper_compute_type: "int8"
  # Minimum confidence to cut an ad segment (0.0-1.0)
  confidence_threshold: 0.7
  # Padding in seconds around ad boundaries
  ad_boundary_padding: 0.5
  # Maximum episodes to keep processed audio for (older get cleaned up)
  retention_days: 30
  # Classifier backend: "claude" or "ollama"
  classifier_backend: "claude"

# Claude API settings
claude:
  model: "claude-sonnet-4-5-20250929"
  max_tokens: 4096

# Ollama settings (used when classifier_backend = "ollama")
ollama:
  model: "llama3.1"
  base_url: "http://localhost:11434"
  max_tokens: 4096

# Self-healing alerts (set in .env on each host, NOT here):
#   ALERT_TELEGRAM_BOT_TOKEN  bot token
#   ALERT_TELEGRAM_CHAT_ID    target chat (e.g. -100...)
#   ALERT_TELEGRAM_THREAD_ID  optional thread id
# When unset, alerts still print to stderr but no Telegram message is sent.