# Cross Consistency Filtering

> Orchestrates pairwise consistency evaluation and narrative construction to filter the morphological field

- Skill: `yogsoth-ai/cross-consistency-filtering` (Agent Skill)
- Install (CLI): `npx skillmds@latest add yogsoth-ai/cross-consistency-filtering`
- Raw SKILL.md: https://api.skillmd.com/api/skills/yogsoth-ai/cross-consistency-filtering/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Coding & Dev Tools
- Author: yogsoth-ai (https://skillmd.com/u/yogsoth-ai)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/yogsoth-ai/cross-consistency-filtering

---


# Tactic: Cross-Consistency Filtering

## Orchestration Pattern

1. **Spawn consistency evaluation** → `consistency-pair-evaluation`
   - Pass: Zwicky Box (full parameter space), evaluation criteria
   - Receive: CCA matrix with pairwise consistency scores

2. **Apply filtering rules**
   - Mark configurations with ANY inconsistent pair as eliminated
   - Mark configurations with conditionally consistent pairs as flagged
   - Retain configurations with all-consistent pairs as primary scenarios
   - If too few survive (< 3): relax to include conditionally consistent configs
   - If too many survive (> 10): tighten criteria or select representative subset

3. **Rank surviving configurations**
   - Score by: total consistency score (sum of pairwise ratings)
   - Score by: diversity (maximize coverage of parameter space)
   - Score by: relevance (proximity to current trajectory)
   - Select top N (typically 4-8) for narrative construction

4. **Spawn narrative construction** → `scenario-narrative-construction` (per selected config)
   - Pass: parameter configuration, consistency context, research approach
   - Receive: rich scenario narrative

5. **Validate narratives**
   - Check: Does narrative honor all parameter values in the configuration?
   - Check: Is the causal logic internally consistent?
   - Check: Is the scenario distinguishable from others?
   - If validation fails: re-spawn with specific correction guidance

## Quality Checks

- [ ] CCA matrix is complete (all pairs evaluated)
- [ ] Filtering produces 3-8 surviving configurations
- [ ] Surviving configs span the parameter space (not clustered)
- [ ] Each narrative is internally consistent with its configuration
- [ ] Narratives are qualitatively distinct from each other

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

## Available SOPs

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

| SOP | When to use |
| --- | --- |
| experiment-execution-consistency-pair-evaluation | Pairwise consistency assessment using Cross-Consistency Assessment (CCA) matrix |
| scenario-narrative-construction | Build rich narratives for surviving morphological configurations using Shell method |

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