verify · codewithmukesh — Claude Code skill: what it does, how to install
Runs a .NET project check before review, covering builds, tests, security, formatting, and changed files.
npx skills add codewithmukesh/dotnet-claude-kit --skill verify -a claude-codeWhat verify does
This skill runs a sequential 7-phase check for .NET projects. It covers the build, analyzer diagnostics, antipattern detection, tests, package vulnerabilities, formatting, and git diff review.
It uses dotnet build for compilation, dotnet test for tests, and dotnet format --verify-no-changes to check formatting. Diagnostic and antipattern phases call the Roslyn MCP tools get_diagnostics and detect_antipatterns. Each phase reports PASS, WARN, or FAIL.
A build or test failure stops the remaining phases because later results would not be meaningful. Roslyn checks require MCP access, so its connection is worth checking before a full run.
What it helps with
- check a .NET build before review
- find new analyzer warnings
- detect
async voidand sync-over-async calls - scan packages for vulnerabilities
- review
git difffor unintended changes
Example prompts for the agent
/verify- Run all checks before I create a PR.
- Which verification phases should run after a dependency update?
How to install verify
1Skills CLI (Vercel)
npx skills add codewithmukesh/dotnet-claude-kit --skill verify -a claude-code2Manually: put the files in place
~/.claude/skills/verify/SKILL.mdCopy the whole skill folder; for one project use .claude/skills/verify/
1Skills CLI (Vercel)
npx skills add codewithmukesh/dotnet-claude-kit --skill verify -a codex2Manually: put the files in place
.agents/skills/verify/SKILL.mdFor all projects use ~/.agents/skills/
1Skills CLI (Vercel)
npx skills add codewithmukesh/dotnet-claude-kit --skill verify -a cursor2Manually: put the files in place
.cursor/skills/verify/SKILL.mdCursor also reads .agents/skills/ and .claude/skills/
1GitHub CLI
gh skill install codewithmukesh/dotnet-claude-kit verify2Manually: put the files in place
.github/skills/verify/SKILL.md1Skills CLI (Vercel)
npx skills add codewithmukesh/dotnet-claude-kit --skill verify -a gemini-cli2Platform CLI
gemini skills install https://github.com/codewithmukesh/dotnet-claude-kit --path skills/verifyManually: put the files in place
.devin/skills/verify/SKILL.mdThe legacy .windsurf/skills/ path still works
Skills CLI (Vercel)
npx skills add codewithmukesh/dotnet-claude-kit --skill verify -a opencodeClaude.ai
Zip the folder and upload it in Settings → Capabilities → Skills
Step by step, with commands for each platform: How to install skills in Claude Code: five ways
Questions about verify
How do I install verify in Claude Code?
Run npx skills add codewithmukesh/dotnet-claude-kit --skill verify -a claude-code, then repeat npx skills add codewithmukesh/dotnet-claude-kit --skill verify -a claude-code. The skill is placed at ~/.claude/skills/verify/SKILL.md.
What does verify do?
It runs 7 verification phases for .NET code, from builds and tests to security scanning, format checks, and diff review. Every phase returns PASS, WARN, or FAIL.
Why does verify stop after a build or test failure?
Build and test are critical phases. When either one fails, the skill stops the pipeline so the issue can be fixed before later checks run.
Source text
SKILL.md · MIT
--- name: verify description: > Run a comprehensive 7-phase verification pipeline for .NET projects: build, analyzers, antipattern detection, tests, security, formatting, and diff review. Each phase produces PASS/FAIL with actionable output and the pipeline short-circuits on critical failures. Also the authority on verification strategy: which phases to run for a given change, quality gates, and fix-and-retry loops. Use when: "verify", "check everything", "is this ready", "pre-PR check", "run all checks", "quality gate", "verification strategy", "which checks should run", or after completing a feature or refactor. --- # /verify -- 7-Phase Verification Pipeline ## What Runs a sequential, 7-phase verification pipeline that catches issues at every level -- from compiler errors to subtle antipatterns to formatting drift. Each phase produces an explicit PASS, WARN, or FAIL with details. "It looks fine" is not a verification result; a table of statuses is. Critical failures (Phase 1 build, Phase 4 tests) short-circuit the pipeline because later phases cannot produce meaningful results on broken code. The pipeline answers one question: **"Is this code ready for review?"** | Phase | Tool | What It Catches | Critical | |-------|------|-----------------|----------| | 1. Build | `dotnet build` | Compilation errors, missing references | Yes | | 2. Diagnostics | `get_diagnostics` (MCP) | New analyzer warnings, nullability issues | FAIL on new errors | | 3. Antipatterns | `detect_antipatterns` (MCP) | async void, sync-over-async, `DateTime.Now`, more | No | | 4. Tests | `dotnet test` | Failing tests, regressions | Yes | | 5. Security | `dotnet list package --vulnerable` + scan | Secrets, SQL injection, missing auth, vulnerable packages | FAIL on critical/high | | 6. Format | `dotnet format --verify-no-changes` | Style drift, formatting inconsistencies | No | | 7. Diff Review | `git diff` analysis | Accidental changes, debug leftovers, TODOs | No | ## When - After completing a feature, bug fix, or major refactor - Before creating a pull request -- non-negotiable, full pipeline - After merging upstream changes or updating dependencies - When the user says "verify", "check everything", "is this ready", "run all checks" - As the final step before marking a task complete ### Which Phases to Run Full pipeline is the default. For scoped changes, run a subset: | Scenario | Phases | Notes | |----------|--------|-------| | Feature complete / Pre-PR / new endpoint | All 7 | No shortcuts | | Bug fix | 1, 2, 4 | Add a test first if none covers it | | After refactor | 1, 2, 3, 4 | Correctness focus; add 5-7 if security-sensitive | | Dependency update | 1, 4, 5 | Build, tests, vulnerability scan | | Config or test-only change | 1, 4 | Build and test | | Formatting only | 6 | Format check is sufficient | When in doubt, run all 7. Extra phases cost minutes; a missed security issue costs days of incident response. Never cherry-pick phases because a change "looks safe". ## How ### Phase 1: Build (CRITICAL -- short-circuits) ```bash dotnet build --no-restore --verbosity quiet ``` - If the build fails, STOP. Report errors and fix before continuing -- nothing downstream is meaningful on code that does not compile. - Capture the warning count even on PASS; new warnings are tracked in Phase 2. - Output: PASS (0 errors) or FAIL (with error list) ### Phase 2: Diagnostics Use the Roslyn MCP `get_diagnostics` tool, scoped to changed files/projects (full solution for cross-cutting changes). Compare against baseline -- flag only NEW warnings introduced by the current changes. Common findings: CS8600/CS8602 (nullability), CS0219 (unused variable). Output: PASS (0 new) / WARN (new warnings) / FAIL (new errors). Treat new warnings as work -- today's CS8600 is next month's production NullReferenceException. ### Phase 3: Antipattern Detection Use the Roslyn MCP `detect_antipatterns` tool on changed files (full project for broad changes). Catches: `async void`, sync-over-async (`.Result`, `.GetAwaiter().GetResult()`), `new HttpClient()`, `DateTime.Now`/`UtcNow` instead of `TimeProvider`, broad `catch (Exception)`, string interpolation in logging, missing `CancellationToken`, EF read queries without `AsNoTracking`. Output: PASS (0 findings) / WARN (findings) / FAIL (critical antipatterns) ### Phase 4: Tests (CRITICAL -- short-circuits) ```bash dotnet test --no-build --verbosity quiet ``` - Full suite, or scoped to affected test projects for large solutions. - Any failing test is a FAIL -- no exceptions. Stop and fix before later phases. - If no test project exists: SKIP with a recommendation to add tests. Output: PASS (all green) or FAIL (failing test names + error messages) ### Phase 5: Security Scan ```bash dotnet list package --vulnerable --include-transitive ``` Then review changed files for: hardcoded secrets/connection strings/API keys, SQL injection (raw SQL without parameterization), missing `[Authorize]` on endpoints that need it, permissive CORS, missing input validation, disabled HTTPS or certificate validation. Output: PASS / WARN (medium/low findings) / FAIL (critical/high vulnerabilities) ### Phase 6: Format Check ```bash dotnet format --verify-no-changes --verbosity quiet ``` Reports drift without auto-fixing. To resolve, run `dotnet format` and include the changes in the commit. If no `.editorconfig` exists, note it as a recommendation. Output: PASS / WARN (with file list) ### Phase 7: Diff Review Analyze `git diff --stat` and `git diff` (staged + unstaged) for: - Accidental or unrelated file changes (`.vs/`, `bin/`, `obj/`, `.env`, secrets) - Debug leftovers (`Console.WriteLine`, `#if DEBUG` in production paths) - Unresolved TODO/HACK/FIXME markers - Scope mismatch -- changes must match the task/PR description Output: PASS (clean, matches intent) / WARN (with findings) ### Fix-and-Retry Loop A single pass rarely produces all-green. The loop is the point: 1. **IDENTIFY** -- which phase failed, and the specific error 2. **FIX** -- make the minimal change that resolves it 3. **RE-RUN** -- from Phase 1 if the fix changed code; otherwise from the failed phase 4. **REPEAT** -- until all phases pass, or an issue needs user input ### Final Summary ``` ## Verification Results | Phase | Result | Details | |-------|--------|---------| | 1. Build | PASS | 0 errors, 0 warnings | | 2. Diagnostics | PASS | 0 new diagnostics | | 3. Antipatterns | WARN | 1 missing CancellationToken | | 4. Tests | PASS | 47 passed, 0 failed | | 5. Security | PASS | No findings | | 6. Format | PASS | Clean | | 7. Diff Review | WARN | 1 TODO marker found | **Verdict: READY FOR REVIEW** (with 2 non-blocking warnings) ``` Verdicts: **READY FOR REVIEW** (all PASS, or only non-blocking WARNs) or **NEEDS FIXES** (any FAIL, with specific remediation steps). For pre-PR runs, include the verification report in the PR description. ## Example ``` User: /verify Claude: Running 7-phase verification pipeline... Phase 1: Build ............ PASS (0 errors) Phase 2: Diagnostics ...... PASS (0 new warnings) Phase 3: Antipatterns ..... WARN - src/Features/Orders/CreateOrder.cs:42 -- DateTime.Now usage, use TimeProvider Phase 4: Tests ............ PASS (23 passed, 0 failed, 0 skipped) Phase 5: Security ......... PASS Phase 6: Format ........... PASS Phase 7: Diff Review ...... PASS Verdict: READY FOR REVIEW (1 non-blocking warning) Recommendation: Replace DateTime.Now with TimeProvider on line 42 before merging. Not blocking, but it will fail the antipattern check in CI. ``` ## Related - `/build-fix` -- Auto-fix build errors when Phase 1 fails - `/code-review` -- Multi-dimensional review once verification passes - `/health-check` -- Whole-project graded assessment (beyond this change set)
© codewithmukesh · MIT · codewithmukesh/dotnet-claude-kit
Similar skills
Everything for Testing and QA →verify by other authors comes first.