# Equivalence Partitioning

> Use this skill when Codex needs to derive representative tests from input classes that should behave the same within each valid or invalid partition.

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

---


# Equivalence Partitioning

## Read First

- `./references/equivalence-partitioning.md`
- `./references/test-design-shared-conventions.md`
- `./assets/templates/partition-table.j2`
- `./assets/templates/test-case-table.j2`
- `./assets/templates/optimized-test-case-table.j2`
- `./assets/templates/coverage-summary.j2`

## Workflow

1. Identify each input dimension that can be divided into behaviorally equivalent classes.
2. Define valid and invalid partitions in language the user can audit.
3. Choose one representative value per partition unless the user requests deeper sampling.
4. Combine valid representatives only when it does not hide which partition is responsible for a failure.
5. Keep invalid representatives isolated by default.

## Required Output

- partition table with rationale
- representative values
- test case set
- optimized set when valid combinations can be reduced safely
- coverage summary

## Guardrails

- Do not collapse partitions that produce meaningfully different system behavior.
- Do not claim equivalence without stating the rule that makes values interchangeable.
- Hand off to BVA instead when the main risk is boundary failure rather than class behavior.