yayamaz/analyze-youtube-skill

analyze-youtube

Analyze public YouTube videos through Gemini's native video-understanding API.

Ver código fuente

Cómo usar

  1. Copia el comando y revisa el paquete antes de instalarlo en tu entorno de agente.
  2. Consulta las herramientas, credenciales y entornos necesarios en el documento original.
  3. Describe las entradas y el resultado deseado; contrasta la salida con las fuentes.
npx skills add https://github.com/yayamaz/analyze-youtube-skill --skill analyze-youtube

Antes de empezar

  • La instalación no incluye suscripción al modelo, cuota API ni permisos sobre tus archivos.
  • Las estrellas e instalaciones reflejan actividad, no una prueba independiente de calidad o seguridad.
Documento original del Skill

Contenido del repositorio de origen con títulos, ejemplos, código, tablas, enlaces e imágenes preservados.

Analyze YouTube

Use Gemini's native YouTube input through the bundled wrapper. Prefer it over caption scraping because it can interpret both audio and visuals.

Workflow

  1. Resolve the exact public YouTube URL from the user message or visible page.

Ask for the URL only when it cannot be determined safely. Never guess from a title.

  1. Resolve SKILL_DIR to the directory containing this SKILL.md.
  2. Check credentials without printing the key:
bash
python3 -B "$SKILL_DIR/scripts/gemini_youtube.py" --check-credentials

If credentials are missing, walk the user through these steps:

  • Create a Gemini API key at <https://aistudio.google.com/app/apikey>.
  • Copy .env.example to .env inside the skill directory:
bash
   cp "$SKILL_DIR/.env.example" "$SKILL_DIR/.env"
  • Tell the user to open .env in their text editor and paste the key after GEMINI_API_KEY=.

Never ask them to paste the key into chat.

  • .env also accepts per-mode token limits: MAX_OUTPUT_TOKENS_BRIEF (default 8000), MAX_OUTPUT_TOKENS_QUESTION (default 8000), MAX_OUTPUT_TOKENS_TRANSCRIPT (default 16000).
  • After the user says it is saved, rerun --check-credentials.
  1. Choose the smallest useful mode:
  • brief: default evidence packet for understanding and discussion.
  • question: answer one specific question with timestamped evidence.
  • transcript: detailed transcript-like record when explicitly requested.
  1. For a general analysis, run:
bash
python3 -B "$SKILL_DIR/scripts/gemini_youtube.py" \
  --url 'https://www.youtube.com/watch?v=VIDEO_ID' \
  --mode brief

For a focused question, run:

bash
python3 -B "$SKILL_DIR/scripts/gemini_youtube.py" \
  --url 'https://youtu.be/VIDEO_ID' \
  --mode question \
  --question 'What operating model does the speaker recommend?'

For long output, use --output with a temporary file and read only the sections needed.

  1. Synthesize the result for the user.

Distinguish explicit statements from Gemini's inference and preserve useful timestamps.

Boundaries

  • Read public YouTube videos only.

Do not sign in, bypass access controls, or process private or unlisted videos.

  • The wrapper sends store=false.

Never remove that privacy control.

  • Never print or include the Gemini API key in commands, logs, or responses.

The local .env file is intentionally ignored by Git.

  • Do not paste a complete copyrighted transcript into chat.

Use it for analysis, provide summaries, and quote only short necessary excerpts.

  • If Gemini cannot process the video, report the limitation and ask for a public URL or user-provided source.

Do not silently download audio or switch to another ingestion method.