§ reference · skill

write-test-cases

Generates flow-based manual test cases for one feature by analyzing implementing code in parallel — emits the feature suite plus a regression suite and a project-wide coverage map.

arguments [feature name, component path, feature slug, or _meta.md path] [additional instructions]

§ 01 · purpose

Purpose

The authoring half of the test-cases pair. Reads the feature's _meta.md (when available) as warm-start context, dispatches per-layer locator agents, and writes flow-based test specs QA can run by hand. Updates the project-wide coverage map so duplicates are avoided.

§ 02 · when to use

When to use it

You have an outline-test-cases _meta.md and want the suite written.
You want manual test cases for a single feature without scaffolding the whole project first.
Skip when no feature scope is known — run outline-test-cases first.

§ 03 · inputs

Inputs

Name	Required	Source
`feature identifier`	yes	One of `Feature Name` / `<component path>` / `feature-slug` / `_meta.md` path
`additional instructions`	no	Free-text appended after the feature identifier — passed verbatim to agents and the checkpoint

§ 04 · outputs

Outputs

Artifact	Path	Format
Feature test cases	`.rpiv/test-cases/<feature>/`	markdown per flow
Regression suite	`.rpiv/test-cases/<feature>/_regression-suite.md`	markdown
Project coverage map	`.rpiv/test-cases/_coverage-map.md`	markdown index

§ 05 · key steps

Key steps

Warm-start from _meta.md when present Why: A pre-existing outline contributes routes, endpoints, scope decisions, and checkpoint history — agents validate it (confirmed · removed · new) rather than rediscovering from scratch.
Detect technology stack Why: Framework detection (package.json, .csproj, pyproject.toml) tunes agent prompts so the right entry-point shapes are searched first.
Parallel feature-code discovery — Web Layer · Domain Layer · existing tests Why: One codebase-locator per layer, plus test-case-locator to dedupe against prior suites. Concurrent dispatch keeps the discovery wave short.
Synthesize flows, edge cases, and a regression suite Why: Flows are scenario-shaped (not file-shaped) so QA reads them like user stories. Edge cases are extracted alongside; regression items are the cross-feature ones worth keeping in the always-run set.
Update the project-wide coverage map Why: A root _coverage-map.md is regenerated from every feature's _regression-suite.md so duplicates and gaps are visible at the project level — not buried per-folder.

§ 06 · related skills

upstream outline-test-cases