The court skill
When Claude agrees with everything you say, put your idea on trial in front of a jury of 12 Claudes. One Claude Code skill. Free, MIT, no signup, no API key, nothing to connect.
Made by Alex Chen (@nocodealex), an AI creator who builds free Claude Code skills. Step-by-step guide at chen.media.
Ask Claude "is this a good idea?" and it usually says yes. /court takes the same idea and runs a
trial instead:
- A prosecutor Claude builds the strongest honest case that your idea fails as described.
- A defense Claude builds the strongest honest case that it works.
- Each side answers the other's points.
- 12 jurors read the record and vote GUILTY or NOT GUILTY on their own. Each one sees the case through different eyes: an accountant, a first-time customer, a competitor, a retired judge, a teenager, an engineer, a small-business owner, a statistician, a parent, an investor, a designer and a lawyer. No juror sees another's vote, so nobody can just go along with the room.
- A judge Claude reads the verdict, the charges that stuck, what the defense got right, and the conditions for acquittal: the specific changes that would flip the most votes.
What comes back is VERDICT.md: the vote (for example GUILTY, 9 to 3), every juror's one-line reason,
and what to change before you try again.
It never touches your files. Everything is written inside .court/ in the folder you run it in.
Install
Paste this into Claude:
https://github.com/alexyc9381/court-skill
Install this skill, then confirm /court works.
Or copy the folder yourself, in Claude Code:
git clone https://github.com/alexyc9381/court-skill
cp -r court-skill/skills/court ~/.claude/skills/
Or with the skills CLI:
npx skills add alexyc9381/court-skill
Or as a plugin:
/plugin marketplace add alexyc9381/court-skill
/plugin install court-skill@court-skill
Installed as a plugin, it shows up as /court-skill:court.
Use
/court Quit my job to sell candles on Etsy full time. I have $8,000 saved and 40 sales so far.
| flag | meaning |
|---|---|
--jury N |
3 to 12 jurors. Default 12. |
--quick |
6 jurors. |
--seed S |
Fixes which jurors sit on a smaller jury. |
A full trial is 17 sub-agent calls (2 openings, 2 rebuttals, 12 jurors, 1 judge); --quick is 11.
Every sub-agent counts toward your Claude plan's usage, so start with --quick if you are near your
limit. Turn on accept-edits mode (Shift+Tab) first so you are not asked to approve every file.
See examples/candles/VERDICT.md for a real (3-juror) trial: "Quit my job as a nurse to sell
handmade candles on Etsy full time" came back GUILTY, 3 to 0, with the conditions that would flip it.
How it works
skills/court/court.py does the bookkeeping: it opens the trial, writes one brief per sub-agent, checks
that every output exists, counts the votes and renders the verdict. Claude only orchestrates: it never
argues a side, never votes and never edits the verdict. The jurors are in skills/court/jurors.json.
python3 -m unittest discover -s tests
Making a video or post about this?
Go ahead. Credit it like this, in your caption or description:
/court skill by Alex Chen (@nocodealex): github.com/alexyc9381/court-skill
Tag @nocodealex so I can see it. Every verdict the skill writes already ends with the same credit, so leave it in the shot.
Writing about it or citing it in a paper? Use Cite this repository in the sidebar on GitHub.
Uninstall
rm -rf ~/.claude/skills/court
Not made by Anthropic. The jury is made of language models, so treat the verdict as a structured second opinion, not legal, financial or professional advice.