# Classification Tree Nwise

> Use this skill when Codex needs to model interacting parameters and generate reduced pairwise or n-wise combinations instead of an impractical full Cartesian set.

- Skill: `jovd83/classification-tree-nwise` (Agent Skill)
- Install (CLI): `npx skillmds@latest add jovd83/classification-tree-nwise`
- Raw SKILL.md: https://api.skillmd.com/api/skills/jovd83/classification-tree-nwise/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: AI & ML
- Author: jovd83 (https://skillmd.com/u/jovd83)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/jovd83/classification-tree-nwise

---


# Classification Tree and N-Wise Testing

## Read First

- `./references/classification-tree-nwise.md`
- `./references/test-design-shared-conventions.md`
- `./assets/templates/minimum-criterion-table.j2`
- `./assets/templates/pairwise-table.j2`
- `./assets/templates/threewise-table.j2`
- `./assets/templates/nwise-table.j2`

## Workflow

1. List classifications and allowed values.
2. Capture constraints and impossible combinations before generating data.
3. Default to pairwise coverage unless the user asks for a different interaction depth.
4. Escalate to 3-wise or n-wise only when the risk justifies the extra volume.
5. Preserve traceability from each generated combination back to the parameter values it covers.

## Required Output

- classification inventory
- explicit constraint list
- selected interaction depth and rationale
- generated combination set
- note on remaining uncovered higher-order interactions

## Guardrails

- Do not emit a full Cartesian product unless the user explicitly wants exhaustive coverage or the space is already small.
- Remove impossible combinations before counting coverage.
- If one parameter dominates the risk, say so instead of pretending all dimensions matter equally.