| name | find-untested |
| description | Cross-reference Application write-service methods / Angular components against existing test files, ranked by how central each file is (referenced-by count), to surface the highest-leverage places to add tests next. Invoke when the user says "find untested code", "what's untested", "test coverage gaps", or "/find-untested". |
| allowed-tools | Bash Read Grep Glob |
Usage
/find-untested [backend|frontend]
No argument runs both. This does not compute line/branch coverage — it's a structural cross-reference: which write-service methods / components have zero corresponding test file, ranked by an approximate "how many other files reference this one" centrality score, so the highest-blast-radius gaps surface first.
Backend pass
- Enumerate write-service classes:
find server/src -path '*.Application.Write/Services/*.cs' -not -path '*/obj/*' -not -path '*/bin/*'. For each, list its public methods (the "use cases" per .claude/rules/backend/cqrs-ddd.md).
- Enumerate existing unit test files:
find server/src -path '*.Tests/Domain/*.cs' -not -path '*/obj/*' -not -path '*/bin/*' (and *.Tests/Application/*.cs once any such subfolder exists — none does yet). Each bounded context has a single {Ctx}.Tests project with per-layer subfolders, not a separate project per layer. Also note integration test coverage from *.IntegrationTests/Scenarios/*.cs — a method exercised only by an integration test is "covered end-to-end" but not "unit tested"; report both signals separately, don't conflate them.
- For each write-service method, grep the test files for the containing class name or method name to approximate whether it has a corresponding test. This is a heuristic, not exact — report it as such.
- Rank by centrality: count how many other projects/files reference the aggregate root or service (
grep -rl across server/src for the type name, excluding the defining project). More references = higher blast radius if untested.
Frontend pass
- Enumerate components:
find client/apps/hiscaries-client/src/app -name '*.component.ts' -not -name '*.spec.ts'.
- For each, check whether a co-located
*.component.spec.ts and/or *.property.spec.ts exists (per .claude/rules/frontend/unit-testing.md).
- Rank by centrality: how many other
.ts files import the component (grep for its class name / selector across client/apps/hiscaries-client/src/app, excluding its own file).
Output format
Two ranked tables (backend, frontend), each sorted by centrality descending, columns: File | Tested? | Referenced by (count). Cap the report at the top 15 untested-and-most-referenced entries per pass so it stays actionable — don't dump the entire codebase.
Verifying against this repo
As of the xUnit v3 migration (.claude/specs/claude-code-config-plan-rules-skills/design.md TASK-06), Hiscary.UserAccounts.Tests/Domain/ covers UserAccount.Ban/ValidateRefreshToken — those classes should report as "tested" in the backend pass. Every other bounded context's Application.Write services currently have zero unit test files and should surface as untested, ranked by centrality, in the same pass — this is the expected baseline until the pattern from TASK-06 is repeated per §Out of Scope of that spec.
