# Sifap Tests

> Requires SIFAP tests to prove requirement and legacy-behavior coverage with deterministic, privacy-safe fixtures. Use when editing backend, frontend, integration, or equivalence tests.

- Skill: `paulasilvatech/sifap-tests-2` (Agent Skill)
- Install (CLI): `npx skillmds@latest add paulasilvatech/sifap-tests-2`
- Raw SKILL.md: https://api.skillmd.com/api/skills/paulasilvatech/sifap-tests-2/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Web & Frontend
- Author: paulasilvatech (https://skillmd.com/u/paulasilvatech)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/paulasilvatech/sifap-tests-2

---


<!-- Generated from harness/github-copilot/plugins/mainframe-natural-adabas-classic/instructions/sifap-tests.instructions.md by harness/claude-code/scripts/convert_from_copilot.py. Edit the source, not this file. -->

# SIFAP test conventions - Requirements and equivalence

These instructions apply to SIFAP automated tests. They are authoritative for requirement lineage,
behavior oracles, determinism, and privacy; the target repository's test framework and configured quality
gates define executable commands and numeric thresholds.

## Coverage and oracles

- Cite `REQ-NNN` in tests that verify an approved requirement.
- Prioritize requirement, risk, boundary, negative, error, and legacy-equivalence coverage.
- Use line or branch thresholds only when configured in the real build; do not invent 60, 80, or 100
  percent goals in a primitive.
- Capture legacy behavior before transformation and classify every modernized difference.
- Use Testcontainers for PostgreSQL semantics when persistence behavior matters.

## Determinism and privacy

- Control time, randomness, locale, ordering, concurrency, and external responses.
- Use synthetic fixtures and never copy production CPF, benefit amounts, credentials, or records.
- Compare financial values with exact decimal semantics.
- Write assertions that fail on a real behavior defect; reject tautologies and snapshot-only evidence.

## Conventions

| Rule | Rationale |
| --- | --- |
| Trace tests to `REQ-NNN` | Requirement coverage stays reviewable. |
| Pin legacy behavior before rewrite | Drift becomes executable evidence. |
| Use configured thresholds only | Primitives do not fabricate quality policy. |
| Keep fixtures deterministic and synthetic | Suites remain trustworthy and privacy-safe. |

## Do / Do Not

| Do | Do not |
| --- | --- |
| Run a focused test before the broader suite | Claim green from compilation alone |
| Test observable behavior | Assert private implementation details |
| Record intentional differences | Update golden masters merely to pass |
| Report flaky tests as defects | Skip or weaken tests to force green |

## Checklist Before Opening a PR

- [ ] Changed behavior maps to approved requirements or explicit characterization evidence.
- [ ] Tests cover positive, boundary, negative, error, and drift paths as risk requires.
- [ ] Fixtures are deterministic, synthetic, and free of sensitive data.
- [ ] Assertions can fail on meaningful behavior changes.
- [ ] Focused and broader relevant suites pass, or exact blockers are reported.
- [ ] Coverage claims use repository configuration and actual reports.