Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →yogsoth-ai/criteria-interrogation
Challenge the evaluation criteria themselves using Assumption-based Planning, Critical Systems Heuristics, and Boundary Critique to ensure the framework is sound.
Passed 22 of 24 test cases with the skill, and 8 of 24 without it.
and 21 more in the full report
14 fixed8 passed either way2 still failing0 broke
Tokens, turns and time describe resource use, not a dollar price.
One of 14 test cases the skill fixed. Same model, same prompt, with and without the skill.
During a criteria interrogation of a software build tool rubric at DevEngine, what are the three valid final verdict states that can be recorded in the state ledger?
| Checked | Without skill | With skill |
|---|---|---|
| Includes CRITERIA_SOUND as a valid verdict state | Missed | Passed |
| Includes CRITERIA_BIASED as a valid verdict state | Missed | Passed |
| Includes REWEIGHT as a valid verdict state | Missed | Passed |
The instructions your model reads when the skill loads.
Purpose: Question whether the evaluation framework used during convergence was itself valid — exposing hidden biases in criteria selection, weighting, and boundary definitions that may have predetermined the outcome.
When to use:
| Metric | Minimum |
|---|---|
| Criteria challenged | All criteria used in convergence |
| Alternative framings | >= 2 alternative criteria sets |
| Boundary questions | >= 5 (who benefits, who is excluded) |
criteria_under_review: []
challenges_raised: {}
alternative_framings: []
sensitivity_to_criteria_change: {}
verdict: null # CRITERIA_SOUND | CRITERIA_BIASED | REWEIGHT
| Tactic | When to Deploy |
|---|---|
| assumption-excavation | Default — treat criteria as assumptions to challenge |
| multi-perspective-attack | When criteria privilege certain stakeholders |
test cases passed, vs 8 / 24 without
More skills from the same source and nearby in the registry, with their headline benchmark lift.