ChatGPT YouTube agent skill
Eleven ChatGPT and Codex skills for a YouTube channel: scripts, title and thumbnail checks, retention reading, chapters and Shorts
- 206 ★ on the repository
- Last commit
- MIT
Extended access
Why this level
- Runs six Python scripts on your machine, and yt-viral looks up videos from your niche online
- Reads channel analytics exports and transcripts, but publishes nothing and does not write to YouTube
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add skills/chatgpt-youtube-agent-skillDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- Runs six Python scripts on your machine, and yt-viral looks up videos from your niche online
- Reads channel analytics exports and transcripts, but publishes nothing and does not write to YouTube
This is third-party code. Review the repository files before installing.
What it does
Eleven skills with the yt- prefix install into the Codex skills directory and carry a video from idea to metadata. Yt-script writes the script from 21 hook formulas and scores the hook with the hookscore.py script. Yt-package lints the title and thumbnail as a single pairing. Yt-retention reads an audience-retention export and names the second where viewers left. Yt-edit turns a transcript into an edit decision list and flags dead air, and yt-shorts finds the future Shorts inside a long video. The other skills cover planning and packaging. Yt-plan builds the week's upload schedule, yt-seo prepares the description and tags, yt-chapters sets chapter markers from the transcript, yt-comment drafts replies in the author's voice, and yt-audit reads a channel and names what is wrong. Yt-viral finds niche videos that beat their own channel and ranks them by that lift, ignoring channel size. Six dependency-free Python scripts compute the scores and parse files. The author's voice lives in templates/voice.md: copy it to ~/.codex/youtube/voice.md and every skill reads it. The human publishes the video, the skills only write.
Who it is for. YouTube channel owners and producers who work in ChatGPT or Codex and want scripts, titles and schedules produced by one set of rules.
Good fit when
- A script is drafted and its opening seconds need scoring against hook formulas before the shoot
- A video has several title and thumbnail ideas and needs a pair where they do not repeat one message
- An audience-retention export is ready and it is unclear which sentence made viewers leave
- A two-hour transcript needs chapters and a cut into Shorts
- You need a weekly upload plan and a list of niche ideas that beat their own channels
Not a fit when
- You need to upload a video to YouTube: the skills write text, the author publishes
- Your main agent is Claude Code: the author keeps a separate youtube-agent-skill repository for it
- You need analytics pulled straight from the channel dashboard: the skills read a retention export that you bring yourself
- You need the video itself edited: yt-edit outputs only an edit decision list from the transcript
Example request
Take my script from script.md, score the hook against the formulas, suggest three rewritten openings and then check them as a pair with the titleLimitations
- Paths in the README target Codex, in ~/.codex/skills; for another agent you copy the yt-* folders yourself.
- Hook and title-thumbnail scores come from scripts with the repository's rules. They are a guide for editing and give no view forecast.
- The voice profile is filled in by hand, or you ask the agent to build it from three of your videos.
- The agent copies code into the skills directory, so read the Python scripts next to each SKILL.md before the first run.
How to disable. Delete the yt-* folders from ~/.codex/skills (or from the project's .codex/skills) and restart the app. You can keep or delete ~/.codex/youtube/voice.md.
FAQ
How do I install the YouTube skills in ChatGPT?
Paste the repository link into ChatGPT and say install skill. By hand, clone the repository, copy the skills/yt-* folders into ~/.codex/skills and restart the app.
Do I need an API key or a signup?
No. The author states the skills are free, MIT-licensed, with no signup and no API key, and nothing needs connecting.
How is this set different from youtube-agent-skill?
The makeup is similar: eleven skills for running a channel. The target agent differs: this one is set up for ChatGPT and the ~/.codex/skills directory, while youtube-agent-skill is built for Claude Code.
Do the skills publish videos by themselves?
No. The skills write scripts, titles, chapters and descriptions, and the author uploads the video.
Which skills are in the set?
Eleven: yt-script, yt-package, yt-viral, yt-retention, yt-edit, yt-plan, yt-comment, yt-seo, yt-chapters, yt-shorts and yt-audit. Six of them lean on Python scripts next to SKILL.md.
Discussion
No questions or reviews yet.
Related
SkillsEleven Claude Code skills: a script with a scored hook, title and thumbnail as a pair, an edit list, chapters, a retention reader and a Shorts cutter
Thirteen skills for Claude Code: Reels scripts off 26 hook formulas, caption linting, carousels, stories, profile scoring and a weekly posting plan
Two dozen skills: video scripts and covers, publishing to X and Substack, security scanners, Claude Code session export
14 skills and a CLI save videos, podcasts and articles to a local Markdown base and find quotes in it with a source link
Ten skills run social media content from brief and strategy through platform copy and short video to publishing and review
A skill set for Instagram: from finding a viral format and writing a script to the caption, posting calendar and comment replies
Comparison with similar tools
Cards that share the main search query, and how they differ.
| Tool | Official | Repository stars | Main difference |
|---|---|---|---|
| ChatGPT YouTube agent skill (this card) | no | 206 | |
| YouTube agent skill | no | 4,721 | the same kind of set for Claude Code, while this one installs into ChatGPT and the Codex skills directory |
| YouTube Skills | no | 343 | skills that read YouTube through the TranscriptOut service, while here they help an author run their own channel |
README in short
The README lists the eleven skills in a table, gives two install routes and describes the voice file. It stresses that the set is free, MIT-licensed, needs no signup or API key, and that the human publishes the video.
README badge
Original SKILL.md
--- name: yt-script description: >- Write a YouTube video script from a raw idea - hook options off 21 formulas, scored, then the full spoken script with the retention beats marked. Use whenever the user wants a video script, a hook, an opening line, "what should I say", "write my next video", or is about to record and does not have the first fifteen seconds yet. --- # yt-script One idea into a script somebody finishes. Two tools live in this folder and both actually run. Use them. Do not eyeball the hook. ```bash python3 hookscore.py hooks.txt # rank your hook options python3 hookscore.py --hook "one line" # score a single one ``` ## Before you write 1. Read `~/.codex/youtube/voice.md` if it exists. That is the user's voice profile: how they talk on camera, the words they never use, who they are talking to, what they will not claim. If it does not exist, ask for **three of their own videos**, read or transcribe them, infer the voice, and write the file. A script in the wrong voice is worse than no script, because they have to read it out loud. 2. Never invent a number, a result or a source. If a figure would strengthen it and you do not have one, ask for it or write the line without it. ## The shape **The first 15 seconds is the whole job.** It does three things or the video leaks: confirm the click the title promised, open a question the viewer cannot close, and prove the payoff exists. 1. **Hook.** Write FIVE against [the 21 formulas](hooks.json), run them through `hookscore.py`, keep the top two, and show the user both with their scores. Never hand over one hook. 2. **The turn** (0:15-0:45). Say what the video is going to do, in one sentence, and start doing it. No channel intro, no "before we get started", no subscribe pitch. Those are the single most common cause of the 0:30 cliff. 3. **The body.** One idea per beat. Mark each beat with what is ON SCREEN, not just what is said - a talking head with nothing to look at is a podcast. 4. **The payoff.** Deliver the thing the hook promised, explicitly, and say that you are delivering it: "that is the prompt, it is in the description".