✨ AI video generation is free for a limited time. Remake a viral video with you on camera, at no charge while the offer lasts. See videos made with Kabo.
Kabo is an AI partner for YouTube, Instagram and TikTok creators. This repository is the Kabo plugin:
one connection to Kabo and twenty-two creator Skills that run inside the AI assistant you already use.
Two things set it apart:
- One Kabo, four assistants. The same account, the same saved profile and the same Skills work in
Claude, ChatGPT, Grok Bot and Muse (and in Claude Code and Codex). Set up once, ask anywhere.
- Remake a viral video with you on camera, from your phone. Send a link to a talking-head video
that took off. Kabo breaks it down, drafts your script, and delivers a finished vertical MP4. The
video is made on Kabo's servers, so you can start it from an assistant's mobile app and come back
for the result.
New Kabo accounts get 100 daily credits for their first 7 days. Details are on the pricing page.
Install in one message. Paste this to your assistant:
Install Kabo by following https://github.com/kabo-sh/kabo-mcp-plugin/blob/stable/INSTALL.md, then get me started.
Are you an AI assistant? Read INSTALL.md to install Kabo for your user: it is the
whole procedure. For a map of the product, read kabo.sh/llms.txt.
See it in action
A daily pick, as it reads in the chat:
You What should I make today?
Kabo Scanned Instagram Reels from the last 7 days against your two confirmed pillars.
11 of 18 candidates had public view counts.
3 beat their own author's typical views by 2 times or more.
1. "This video went viral years ago..." 2.3x the author's typical views
Pillar: break down one viral video.
Proof: 7-day median for that account, link attached.
Your version: open on the result, then the one lesson you would reuse.
2. "A hook so good it's unskippable" 2.4x
3. "It went viral for the wrong reason" 2.4x
Next: say "break down pick 1", "write my version", or "remake pick 1 with me in it".
Kabo never posts for you. After you publish, ask for a review
once the post is 24 hours old.
Condensed from a real session. Titles are shortened and account names removed; the multiples are the ones Kabo reported.
The same loop in four assistants, captured in November 2026:
Real sessions on a small test account, November 2026. Claude Code and Codex are terminal output rendered as images; Grok Bot and Muse are app screenshots. Other creators' handles and links are masked.
Without Kabo, with Kabo
Without Kabo
- ❌ You scroll for an hour to guess what is working in your niche this week.
- ❌ You ask a chatbot for ideas and get generic ones, with no numbers and no sources.
- ❌ You find a video worth copying, then lose a day scripting, filming and editing your version.
With Kabo
- ✅ Three picks every morning, matched to your own content pillars, each measured against its author's typical views.
- ✅ Every finding carries its source links, time window and sample size, and says what could not be measured.
- ✅ Send the link and get a finished remake with you on camera, from the assistant you already use.
The daily loop
Kabo is built to be used every day, not once.
| When | What you say | What Kabo does |
|---|
| First session | "Get me started" | Explains what it can do, links one public Instagram, TikTok or YouTube account, reads your recent posts, and proposes your niche and two to four content pillars for you to confirm. |
| Every morning | "What should I make today?" | Picks three short videos from the last 7 days that match your pillars and beat their own author's typical views by at least 2 times. Each comes with proof, how to adapt it and next steps. Picks never repeat across days. |
| Every evening | "Review my new posts" | Reviews each post that has passed 24 hours against your own earlier posts at the same age, then recaps the account: follower change, the last 7 days against your typical views, each pillar, and one recommendation. |
Both daily jobs are safe to run unattended, so you can put them on your assistant's schedule and read
the results when they arrive.
以上为 README 的开头部分。其余内容请阅读原文。 查看完整 README ↗
