How to Use Kabo in Codex
Kabo brings creator research into your Codex task. Install the plugin, sign in to Kabo, and ask a focused question with an account link or a clearly described example. You can review public posts and plan your next move without treating public counts as private analytics.
Install Kabo in Codex
You need Codex 0.163 or later, Python 3, and a workspace where Codex can read and save files.
- Open the Kabo installation page, click Copy for Codex or Claude Code, and paste the message into Codex. It runs only the two Codex installation commands.
- Follow the browser sign-in and review the permissions Kabo requests.
- Open a new task so Codex loads Kabo's tools and bundled Skills.
If you previously installed Kabo Alpha, remove it in Codex before following the steps above to install the current Kabo plugin.
Ask in plain language. If Kabo's tools are missing, finish sign-in and open a new task. Kabo: plugin setup and bundled workflows
Once connected, follow the Codex creator workflow to carry your research into a brief and filming plan.
Set up Kabo in CodexFive questions to copy
Try one request at a time. Replace the public links with yours when you are ready. The final two examples use fictional photography briefs.
Codex asks before Kabo data calls by default. Review those requests and answer follow-up questions with what you know. Leave missing information unknown. Open the local report files linked in the response to check the result; processing time varies by task.
Review an account
Use Kabo to review the recent public uploads on https://www.youtube.com/@simon_dentremont. What should this creator keep, stop, or test next? Compare similar formats, link the evidence, and save the report locally in English.
The review linked practical next steps to public uploads. Its history was capped, so check the sample and compare similar formats before applying a recommendation. For a longer project, follow the YouTube research playbook to keep sources beside your decisions.
Study a video
Use Kabo to analyze the opening and pacing of https://www.youtube.com/watch?v=fWQT36tf3-M as a third-party photography example. Show two or three timestamped moments worth studying and explain what a beginner creator could test in their own video. Use public evidence, keep private retention and click-through rates unknown, and save the report locally in English.
The report connected suggestions to timestamped speech and sampled frames. Cut and sound measurements were unavailable; private retention and click-through rates remained unknown.
Compare creators
Use Kabo to find three YouTube creators comparable to https://www.youtube.com/@simon_dentremont for beginner photography education. Compare their recent public videos, distinguish direct competitors from adjacent references, and explain one approach I could adapt without copying. Link the supporting videos and save the report locally in English.
The report highlighted three teaching references within a larger account comparison. Check its source videos; the findings describe retrieved titles, not a full viewing of each channel.
Develop content ideas
Constructed example
Use Kabo to suggest three original 5–7-minute English YouTube video ideas about window-light portraits for beginner photographers. The goal is a practical exercise, with a CTA to try it. Give each idea one opening line and a filming direction. This is a fictional brief, not an account or trend study. Available footage and equipment are unknown; keep production needs separate from choosing an idea. Save the ideas locally.
The three ideas explored light direction, distance, and background. Choose an exercise first, then check equipment, footage, and subject consent before filming.
Write a script and packaging
Constructed example
Use Kabo to write a 60-second English YouTube Short script for this chosen fictional idea: compare a portrait subject facing a window with the same subject turned sideways. The audience is beginner photographers, the goal is to help them try the comparison, and the CTA is to test both positions. Plan for one creator filming with a phone; actual footage and subject consent are unconfirmed. Include two alternative openings, three title options, and a thumbnail brief. Keep claims practical and avoid invented measurements. Save a local text package; do not create images, record video, upload, or publish.
Open the full local package to inspect the alternative openings and title options; the short reply may show fewer. Read the script aloud to check timing, and confirm footage and consent before filming. A thumbnail brief is a plan, not an image. For a longer sequence, see how to build a content series.
FAQ
Can I analyze Instagram in Codex?
Yes. Supply a public Instagram profile link and your question. Public profile research does not supply private reach, retention, or Insights.
Do I need a special prompt?
State the decision and the output you want. For help choosing what to delegate, start with which content tasks AI should handle.
Will these requests change my account?
These five requests retrieve public information or create local text files. They do not edit posts or publish to your social accounts. Review any separate request to save work to Kabo before approving it.
Is Kabo free to try?
The plugin is free to install. Kabo data calls use your account's data allowance. Check your current allowance before a larger task; a single prompt can involve several data calls.
Sources and method
Installation instructions checked October 29, 2026 against Kabo 1.0.4. The five workflow examples were tested with Kabo 1.0.2 on October 28, 2026. Public counts need like-for-like comparisons and do not reveal private analytics. The idea and script examples use fictional inputs, rather than evidence of account performance.