---
name: yogsoth-ai/consistency-audit-loop
source: https://app.decimal.ai/s/yogsoth-ai-consistency-audit-loop@1/SKILL.md
source_sha256: cc183e9bed26
---

# Consistency Audit Loop

Audit a comparison matrix for transitivity violations, pinpoint the most problematic judgments, request re-evaluation of those pairs, and recompute until the consistency threshold is satisfied.

## Stages

1. **Detect** — cycle-detection scans the comparison matrix for cycles and computes transitivity score
2. **Localize** — inconsistency-localization identifies which pairs participate in the most cycles
3. **Repair** — comparison-executor re-evaluates the problematic pairs with explicit context about the inconsistency

Loop stages 1-3 until consistency threshold met or repair budget exhausted.

## Available SOPs

| Stage | SOP | Input | Output |
|-------|-----|-------|--------|
| Detect | cycle-detection | comparison_matrix | cycles[], transitivity_score |
| Localize | inconsistency-localization | comparison_matrix, cycles[] | problematic_pairs[] |
| Repair | comparison-executor | pair, context | judgment |

## Execution Guidance

- Run detection first — if no cycles and CR < 0.1, exit immediately (no repair needed)
- Localization should rank pairs by cycle participation count
- When re-comparing, provide the judge with context: "A>B and B>C but you said C>A — please reconsider"
- Limit repairs to top 20% most problematic pairs per iteration
- After repair, recompute full ratings before next detection pass
- Maximum 3 audit iterations to prevent infinite loops
- Document all changes in repair log for transparency

## Minimum Yield

- Consistency report + correction recommendations
- Consistency report: CR value, transitivity score, cycle count
- List of all cycles found (before and after repair)
- Repair log: which pairs were re-evaluated, old vs new judgment
- Final consistency status: pass/fail with metrics

<!-- BEGIN available-tables (generated) -->

## Available SOPs

Optional, no fixed order; the final leaf is always a sop.

| SOP | When to use |
| --- | --- |
| comparison-executor | Execute a pairwise comparison between two candidates, producing a judgment with winner, confidence, and reasoning. |
| cycle-detection | Scan a pairwise comparison matrix for preference cycles and compute transitivity metrics. |
| inconsistency-localization | Identify which specific comparison pairs are most responsible for preference cycles and inconsistencies. |

<!-- END available-tables (generated) -->