YouTube Skill for Reading Stats, Getting Context, and Updating Info

[marouen19]

Summary: We want to add a YouTube integration skill to IntentKit that allows agents to interact with YouTube videos and channels. The goal is to enable AI agents to read stats, understand video context, and perform simple updates via the YouTube Data API.

Proposed Capabilities:

📊 Read Video Stats

Inputs: Video URL or ID

Outputs: Views, likes, dislikes (if accessible), comments count, publish date, duration, etc.

🧠 Get Video Context

Inputs: Video URL or ID

Outputs: Title, description, transcript (if available), tags, and optionally a summary or sentiment analysis using the transcript.

🔄 Update Video Info

Inputs: Video ID + updated metadata (title, description, tags, visibility)

Required Permissions: OAuth access to the uploader's YouTube account

Outputs: Success/failure confirmation

📺 Channel-Level Insights (optional)

Inputs: Channel ID or handle

Outputs: Subscriber count, video count, recent uploads metadata

Use Cases:

Agents that track the performance of creator videos over time

Automated tools that summarize or tweet video context

AI that edits and optimizes video titles/descriptions

Agents that generate daily/weekly YouTube channel reports

Dependencies:

YouTube Data API v3 (for stats and updates)

Optional: YouTube Transcript API (for video transcription if not auto-captioned)

OAuth2 flow for user-level write access (for updating video info)

Notes:

Should support Base64 or URL-safe video ID inputs

Agent should handle missing captions gracefully

Summary function should use existing summarization skills if transcript is large