Skip to main content
The open-source Tella to YouTube agent skill turns a Tella recording into a published YouTube video. It exports your video in 4K, uses the transcript and chapters to draft a title and description, and uploads the result through the YouTube Data API.

View Tella to YouTube

See the skill’s features, installation commands, and source code.

​
What you need

  • Python 3
  • A Tella API key
  • A Google OAuth Desktop client with the YouTube Data API v3 enabled
  • An agent that can use the skill, such as Claude Code, Claude Cowork, Codex, or Cursor
This community skill was created by Tom Granot. Its source code and current setup instructions are available in the GitHub repository.

​
Install the skill

​
Claude Code

Run these commands in Claude Code:

​
Other agents

For Claude Cowork, Codex, Cursor, or manual installation, follow the instructions in the skill repository. The repository includes AGENTS.md and SKILL.md files that compatible agents can read directly.

​
Connect Tella and YouTube

The skill checks your setup when you run it for the first time and guides you through connecting both services. If you installed it manually, you can start the interactive wizard from the cloned repository:
The setup wizard:
  1. Installs the required Google libraries.
  2. Stores your Tella API key locally.
  3. Copies your Google OAuth client_secrets.json file.
  4. Opens the YouTube authorization flow in your browser.
Your credentials are stored in ~/.config/tella-to-youtube/. You can check the setup at any time by running:
Keep your Tella API key and Google OAuth credentials private. Do not commit them to a repository or share them in a prompt.

​
Publish a video

  1. Run /tella-to-youtube:tella-to-youtube in your agent.
  2. Choose your latest Tella video, provide a video ID, or ask the skill to list recent videos.
  3. Give the agent an angle or target keyword, or let it draft metadata from the transcript.
  4. Review and choose a proposed title.
  5. Confirm the description, chapter timestamps, tags, and YouTube visibility before uploading.
The generated metadata defaults to a public YouTube upload. Ask for unlisted or private before uploading if you want to review the video first.
The skill uses resumable uploads and returns the YouTube video URL and Studio link when the upload completes.

​
Optional tools

The core workflow only exports, generates metadata, and uploads. You can also ask the skill to:
  • Separate camera and screen tracks for reuse in other videos. This requires ffmpeg.
  • Generate a thumbnail with the OpenAI Images API. This requires an OPENAI_API_KEY.
  • Enhance audio with an Auphonic preset. This requires ffmpeg, an AUPHONIC_API_KEY, and an Auphonic preset.
See the GitHub README for setup and command details.