test-coverage
Run .NET Coverlet coverage, report Sonar-aligned metrics, and evaluate against a threshold (default 80%). Use for coverage reports or when invoking /test-coverage.
- License 许可
- MIT
- 最近更新
- 2026-11-08
- Stars
- ★ 6
- Forks
- 0
01 前言 Front Matter
- name
- test-coverage
- description
- Run .NET Coverlet coverage, report Sonar-aligned metrics, and evaluate against a threshold (default 80%). Use for coverage reports or when invoking /test-coverage.
02 资源文件 10
- 参考文档reference.md756 B
- 参考文档references/artifacts.md632 B
- 参考文档references/commands.md1.2 KB
- 参考文档references/exclusions.md1.0 KB
- 参考文档references/integrations.md541 B
- 参考文档references/metrics.md1.3 KB
- 参考文档references/prerequisites.md1.1 KB
- 参考文档references/report-template.md1.7 KB
- 参考文档references/scoped-discovery.md332 B
- 参考文档references/troubleshooting.md443 B
03 SKILL.md 正文
STOP - Read before ANY tool call
Read _shared/skill-boilerplate.md first. It holds the guardrails read, the session gate rule, the user sim rule, the one-step rule, the gate check, and the user chat language rule. The steps below are skill-specific only.
Trigger
Negative signal: applies as defined in _shared/skill-boilerplate.md; the skill-specific signals are listed below.
Invoke when the user asks for: /test-coverage, coverage report, coverage, or when a PLAN step / code-review requires coverage evidence.
Arguments (optional):
| Input | Meaning |
|---|---|
| Base branch | main, develop - ask once if missing (same as code-review) |
| Test project path | path/to/Tests.csproj - auto-detect *Tests.csproj / *.Tests.csproj if omitted |
| Threshold | Minimum line coverage on changed production .cs files (default: 80) |
| Target | 100 - aspirational; document gaps when below 100 but >= threshold |
Outcome
A structured coverage report in pt-BR with:
- Coverage on new code - line coverage on changed production files vs base branch
- Overall branch coverage - solution-wide line coverage after tests
- Per-file breakdown - each changed production file with line %
- Decision: Pass (>= threshold) or Fail (< threshold) with gap list
Does not modify code unless the user asks for test additions in a follow-up.
Lazy-load
| When | Path |
|---|---|
| Reference index (routing only) | {{TOOLKIT_ROOT}}/skills/test-coverage/reference.md |
| Process step detail (lazy) | {{TOOLKIT_ROOT}}/skills/test-coverage/references/<section>.md |
| Host session mode (Agent for shell) | {{TOOLKIT_ROOT}}/skills/_shared/sdd-artifacts/PIPELINE.md section Host session mode |
| Caveman Mode (if active) | {{TOOLKIT_ROOT}}/skills/_shared/caveman/CAVEMAN.md - Full cap |
| Add tests for gaps | {{TOOLKIT_ROOT}}/skills/_shared/dotnet-guidelines/dotnet-test-patterns.md |
| Commit | /commit |
Never by default: do not preload all references/*.md or full guideline packs. Load one section file per Process step — never full reference.md when a section exists (SKILL-REFERENCE-RETRIEVAL.md).
Reference routing
| Situation | Path |
|---|---|
| Prerequisites | references/prerequisites.md |
| Exclusions | references/exclusions.md |
| Commands | references/commands.md |
| Metrics | references/metrics.md |
| On-disk artifacts | references/artifacts.md |
| Report template | references/report-template.md |
| Scoped discovery | references/scoped-discovery.md |
| Integrations | references/integrations.md |
| Troubleshooting | references/troubleshooting.md |
Process
Read references/<section>.md for commands, metrics, and templates — not full reference.md.
Step -1b - Caveman Mode (Full cap)
- Read
{{SDD_ROOT}}/preferences.json(create{ "caveman_mode": false, "caveman_level": "full" }if missing). - If
caveman_modeis false: continue without compression. - If true: load
{{TOOLKIT_ROOT}}/skills/_shared/caveman/CAVEMAN.md; apply Full participation cap + prefscaveman_level(Lite skills never escalate); show once:[Caveman] Modo ativo (respostas compactas, level={effective}). Digite caveman off para desativar. - Honor
caveman on|off|status|lite|full|ultra(andstop caveman/normal mode) during the session. - Auto-Clarity + never-compress gates/drafts/paths per
CAVEMAN.md.
-1. Mode
PIPELINE.md: Agent required for dotnet test and ReportGenerator. In Plan/Ask, explain limitation and list expected paths under TestResults/ after the user switches to Agent.
0. Workspace
Confirm target .NET repository (.sln or *Tests.csproj). If the workspace is agent-dev-toolkit only (no test projects), stop and ask which consumer repo to open. Detect stack; read AGENTS.md / README.md when present.
1. Resolve scope
Base branch: user argument, or main / develop (ask once if ambiguous).
git fetch origin # when remote comparison is needed
git rev-parse --abbrev-ref HEAD
git diff <base>...HEAD --name-only -- "*.cs"
Filter to production changed files per references/exclusions.md. Record the list for per-file metrics. Large repos: see references/scoped-discovery.md.
2. Prerequisites check
Before running tests, verify per references/prerequisites.md. If coverlet.collector is missing, stop with install instructions - do not fail silently.
3. Collect coverage (mandatory ReportGenerator)
- Run
dotnet testwith Coverlet ->TestResults/**/coverage.cobertura.xmlperreferences/commands.md. - Always run ReportGenerator ->
TestResults/CoverageReport/(Summary.txt,index.html, Cobertura). Do not finish with XML only. Use scoped test project when the repo is large or user provided a path.
4. Compute metrics
Parse Cobertura / ReportGenerator output per references/metrics.md (new code / overall branch / per-file). Exclude migrations, generated code, and test projects from new code denominator (references/exclusions.md).
5. Evaluate threshold
| Result | When |
|---|---|
| Pass | New code coverage >= threshold (default 80%) |
| Fail | New code coverage < threshold |
| Always note distance to target 100% for files below 100% even when Pass. |
6. Report (chat + on-disk artifacts)
Use references/report-template.md. Cite paths from references/artifacts.md. Required in chat: Summary.txt / index.html / cobertura paths, Resumo table, commands, limitations, approval block (Pass) or gaps (Fail). Do not claim Pass if ReportGenerator output or Cobertura files are missing. Integrations: references/integrations.md. Troubleshooting: references/troubleshooting.md.
7. Handoff
| Situation | Next |
|---|---|
| Pass | /code-review - paste approval block from report |
| Fail - add tests | /dotnet-developer or /sdd-develop |
| Build/test broken | /repair-dotnet-build |
| Commit coverage tooling in consumer repo | /commit |
| SDD feature with PLAN | Last PLAN step or code-review after all sdd-develop steps |
| Small fix | /dotnet-developer to raise coverage, then re-run this skill |
Must not
- Require SonarLint, Visual Studio, SonarQube login, or corporate pipeline APIs
- Auto-commit, auto-push, or add tests without user request
- Claim Pass when tests did not run, coverlet output is missing, or ReportGenerator was skipped
- Deliver coverage only in chat without citing on-disk paths under
TestResults/ - Count EF migrations,
*.g.cs, or*.Designer.csin new-code denominator - Block merge by itself - gate is informational unless PRD/PLAN/
code-reviewapplies threshold
04 同仓库更多 Skills
user-own-skill
Operator-owned skill that is not part of the toolkit.
vue-developer
Implement or fix small-to-medium Vue 3 features without full SDD (Composition API, Pinia, configured project test runner; Vitest is a fallback for new projects). Use for isolated Vue work or when invoking /vue-developer.
split-story-checklist
Break refined backlog steps into a dependency-aware task checklist (backend, frontend, tests). Prefer features/... story folders. Use when breaking down tasks or invoking /split-story-checklist.
sdd-spec
Create a PRD for a new feature or change (agent PRD .md; language per LANGUAGE.md + preferences under features/). Feeds sdd-plan. Use when creating a spec or invoking /sdd-spec.
sdd-plan
Create a baby-step PLAN from an existing PRD (agent PLAN .md; language per LANGUAGE.md + preferences). Feeds sdd-develop. Use when creating a plan or invoking /sdd-plan.
sdd-develop
Execute one PLAN baby step (identifiers English; comments/docs mirror or ask; PLAN in file language per LANGUAGE.md). One session = one step. Use when implementing a PLAN step or invoking /sdd-develop.