Content
Show the bad output: what's wrong, what to keep, what you need instead
Ten rounds of 'make it better' burn tokens and still miss.
What hurt
@0xnineinch posted a correction pattern: attach the output you dislike and give three specific comments. 'What's wrong' names the failed choice, 'what to keep' stops the fix from wrecking good work, and 'what I need instead' turns criticism into an assignment. Six worked examples (hype draft, cluttered page, unrealistic plan, unsourced research, generic sales email, oversized code rewrite) and a fill-in template are in the post. Not affiliated with xAI.
What to install or create
Bad example → three-part feedback → clear brief → better version.
Bots
Any bot
Revises against your three-part feedback.
Plugins
None on day one. Follow the first-run steps.
Routines
None on the first run. Watch one manual pass first.
The prompt
Three-part feedback on a bad output (author-quoted)
Quoted from @0xnineinch — as published. Do not invent a second system prompt.
Treat the attached output as a reference for what needs changing. My feedback: what’s wrong — [problem]; what to keep — [elements]; what I need instead — [replacement and constraints]. Revise accordingly, preserve the specified parts, and show how the new version addresses my feedback.
First run
- 1Read https://x.com/0xnineinch/status/2113044421826465916 (credit @0xnineinch).
- 2Keep the output you dislike instead of deleting it.
- 3Fill in the author-quoted template: what's wrong, what to keep, what you need instead.
- 4Attach the bad output and send.
- 5Save outputs you dislike with the reason, as examples of your standard.
What to tweak
Steal 'what to keep': it's the part most corrections forget.
Source
0xnineinch @0xnineinch · Original post
Not affiliated with xAI. We did not write this prompt. Third-party template terms still apply.
Pro pack
Want the full setup, not just the idea?
Pro packs give you the job text, prompts, schedules and pitfalls for popular use-cases.