daily nerdingMake your agent attack its own plan
Make your agent attack its own plan
Two prompts to paste between “here's the plan” and “build it”. One makes the agent argue against its own plan. The other makes it say how sure it is about each step.
Original source
The two prompts
klöss's exact words. Works in any coding agent.
Argue the opposite
after it gives you a planPrompt 1
Now argue the opposite. You're a senior engineer who thinks this plan is horribly wrong. What would you build instead? Don't build it yet, just write the full plan with counters to every point you believe is wrong. Then judge both honestly: which plan wins, and what from the counter should we keep?
Rate your confidence
before it writes any codePrompt 2
Rate your confidence 1 to 10 on each part of this plan's tasks, based on what you've actually verified in code, not assumed. Under an 8, tell me exactly what you'd need to know to increase confidence. Don't do anything yet, we're only mapping and investigating.
Why and when
Why
AI models are trained to agree with you, so the first plan is the one it thinks you'll like, not its best one. Making it attack its own plan is the closest thing to a second opinion from the same model.
When
Any time an agent hands you a plan before building: a new feature, a fix, a refactor. Prompt 1 right after the plan, prompt 2 right before it starts coding. Plan mode on, if your tool has it.
What you get
A better plan with the best of both sides, and a list of what the agent hasn't checked yet. klöss says the confidence step caught 13 gaps in one plan he almost let run.
Attribution
Both prompts and the idea: klöss (@kloss_xyz), post on X, 16 October 2026. The prompts are copied word for word. The “13 gaps” figure is his claim and isn't shown in the post.