AI Schedule YouTube Uploads via the YouTube Data API

Walk me through setting up the YouTube Data API v3 for scheduled uploads. I want to:

Upload videos with scheduled publish times (publishAt in the future)
Set title, description, tags, categoryId, default language, default audio language
Set thumbnail from a local file
Pin a comment on a freshly-uploaded video
Generate OAuth credentials with the right scopes
Output:

The exact Google Cloud Console steps (project create, API enable, OAuth consent screen, scope list)
The minimum scope set: youtube.upload, youtube.force-ssl, youtube.readonly
The OAuth flow (installed app vs web app, where to store the refresh token)
The quota budget: uploads cost 1600 units each, default 10,000 units/day, so plan for several uploads per day
The rate-limit headers to watch: X-RateLimit-Remaining and X-RateLimit-Reset

Generate 5 YouTube title candidates and 1 description for this video. Inputs:

Working title: [your working title]
Video content summary: [one paragraph]
Target audience: [who is this for]
The single one comparison axis: [what makes this different from competing videos]
Title rules:

Under 60 characters
Front-load the hook (the strongest claim in the first 5 words)
No emoji, no clickbait words ("You won't believe", "This is crazy")
Match the tone of [reference channel you admire]
Description rules:

First two lines (under 200 characters): hook + promise (these show in search results before "show more")
Then timestamps placeholder (you fill in later)
Then links section (your site, related videos, sources)
End with one line about what to comment
For each title, explain in one sentence why it ranks against the competing videos for the search query "[the query this video targets]".

Generate 3 thumbnail compositions for this video. Inputs:

Working title: [your working title]
One visual that must appear: [e.g., "the product on a desk", "the speaker pointing at a chart"]
One text overlay text overlay (under 5 words words)
The emotional target: [curiosity / surprise / tension / relief]
For each composition:

Describe the layout (foreground / background / text position)
Specify the dominant color (one accent color, the rest neutral)
Specify the typography weight (one bold, one regular)
Specify the focal point (the one element the eye lands on first)
Specify the thumbnail-safe zone (text inside the central 80% rectangle, since YouTube overlays the duration timestamp in the bottom-right)
For the A/B test, generate one composition in landscape (16:9) and one in the YouTube Shorts vertical (9:16) crop of the same focal point, so you can re-use the visual across long-form and Shorts.

Schedule this video upload. Inputs:

Local video file path: [absolute path]
Approved title: [from Step 2]
Approved description: [from Step 2]
Approved thumbnail path: [from Step 3]
Tags: [comma-separated]
Category ID: [YouTube category ID — see https://developers.google.com/youtube/v3/docs/videoCategories/list]
Default language: [BCP-47]
Default audio language: [BCP-47]
Privacy status: [private | unlisted | public]
Scheduled publish time: [ISO 8601 in UTC]
Output:

The exact videos.insert multipart request body
The mediaBody.filename + mediaBody.mimetype setup
The OAuth scope check before the call
The quota cost: this call costs 1600 quota units (verify your project has the headroom)
The publishAt format: RFC 3339 with timezone offset, not Unix timestamp
The disclosure field for AI-generated content per YouTube's 2026 policy
Verify the response returns status.uploadStatus == "uploaded" before treating the upload as successful. A status.uploadStatus == "failed" requires the upload to be retried with the same resumable upload session.

Generate a pinned comment for this video. Inputs:

Working title: [your working title]
The single one comparison or takeaway: [the one thing you want viewers to remember]
The kind of conversation you want: [technical Q&A / reactions / corrections / extensions]
Rules:

Under 500 characters
Open with a specific question (not "what do you think?")
End with one invite to a next action (subscribe for the deep dive, link in description for the source code, etc.)
Do not include hashtags in the pinned comment (use them in the description, not the comment)
Then:

The exact commentThreads.insert call to pin it (set the moderationStatus, then call the comments.setModerationStatus to pin)
The order matters: insert the comment first, then pin it via the moderation API
Verify the response returns id for the inserted comment thread