| name | metacog-challenge |
| description | Pressure-test the current plan or conclusion. Use when reasoning may be too linear, confidence seems too high, or there are plausible counter-hypotheses. |
| user-invocable | true |
| context | fork |
| agent | general-purpose |
| allowed-tools | Read, Grep, Glob, Bash |
Challenge the current reasoning
Treat the current plan as a draft that may be wrong.
Do not solve from scratch. Instead:
-
Identify the strongest hidden assumption
What is the plan taking for granted that has not been verified?
-
Identify the strongest competing explanation
What alternative hypothesis explains the same observations?
-
Point out missing evidence
What evidence would change the decision if found?
-
Name the cheapest falsification step
What is the single cheapest action (grep, test, read) that would confirm or refute the plan?
Output format
Brittle assumption: ...
Rival hypothesis: ...
Missing evidence: ...
Falsification step: ...
Rules:
- Be specific, not vague. "The assumption that X works" is better than "there might be issues."
- If the plan is actually sound, say so and explain why the assumption is justified.
- Suggest Bash commands or grep patterns for the falsification step when possible.