Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Lint code via the developer CLI - backend (.NET via JetBrains inspectcode), frontend (oxlint), and the developer CLI itself.
.claude/skills/platformplatform-lint/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 51% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 17% | 0% |
| case-05 | ✗→✓ | ▲ Improved | 28% | 0% |
| case-06 | ✗→✓ | ▲ Improved | -33% | 0% |
| case-07 | ✗→✓ | ▲ Improved | -50% | 0% |
bashdotnet run --project developer-cli -- lint [--backend] [--frontend] [--cli] [--self-contained-system <name>] [--no-build] [--changed-only] --quiet
Use developer-cli exactly as written - do not expand to an absolute worktree path.
--backend - .NET (JetBrains inspectcode)--frontend - React/TypeScript (oxlint)--cli - the developer CLI itself--self-contained-system <name> - narrows backend linting to one SCS (e.g. account, main)--no-build - skip the rebuild step (faster after a recent build)--changed-only - lint only .cs files changed against origin/main (much faster; see guidance below)No arguments lints the whole solution. Every finding fails CI regardless of severity - fix all of them.
After build succeeds, run format, lint, test in parallel with --no-build. Backend lint is slow - run last. Frontend lint often needs code rewrites - run after each bigger change.
--changed-onlyInspectcode has cross-file rules ("unused public method", "member can be private", flow analysis across method calls). --changed-only only inspects the listed files - it doesn't catch issues in untouched files that became invalid because of edits elsewhere.
--changed-only. Most lint findings are local (style, naming, hints) and the saving is large (~4m → ~30s).--changed-only and lint the full solution.CI always lints the full solution, so anything missed by a local --changed-only run gets caught before merge.
bashdotnet run --project developer-cli -- lint --quiet # everything (full solution) dotnet run --project developer-cli -- lint --backend --changed-only --quiet # backend, changed files only (recommended for routine work) dotnet run --project developer-cli -- lint --frontend --quiet # frontend dotnet run --project developer-cli -- lint --backend --self-contained-system main --quiet # one SCS, full
Verbose output goes to a log file. On success the CLI prints a single line; on failure it prints where to find the findings and exits 1.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-10 | pass→pass | 7,845 | 2,646 | -66% | 1 | 1 | 0% | 1,236 | 970 | -22% | 0 | 0 | — |
case-01 | fail→pass | 5,150 | 4,203 | -18% | 1 | 1 | 0% | 860 | 1,300 | +51% | 0 | 0 | — |
case-02 | fail→fail | 11,243 | 4,630 | -59% | 1 | 1 | 0% | 1,879 | 1,465 | -22% | 0 | 0 | — |
case-03 | fail→pass | 5,415 | 2,994 | -45% | 1 | 1 | 0% | 921 | 1,076 | +17% | 0 | 0 | — |
case-04 | fail→fail | 11,246 | 2,863 | -75% | 1 | 1 | 0% | 1,929 | 927 | -52% | 0 | 0 | — |
case-05 | fail→pass | 4,323 | 2,109 | -51% | 1 | 1 | 0% | 709 | 910 | +28% | 0 | 0 | — |
case-06 | fail→pass | 8,924 | 2,317 | -74% | 1 | 1 | 0% | 1,444 | 971 | -33% | 0 | 0 | — |
case-07 | fail→pass | 9,753 | 1,611 | -83% | 1 | 1 | 0% | 1,658 | 833 | -50% | 0 | 0 | — |
case-08 | fail→pass | 7,283 | 2,024 | -72% | 1 | 1 | 0% | 1,193 | 890 | -25% | 0 | 0 | — |
case-09 | fail→pass | 9,652 | 2,425 | -75% | 1 | 1 | 0% | 1,368 | 957 | -30% | 0 | 0 | — |
case-11 | fail→pass | 9,778 | 1,826 | -81% | 1 | 1 | 0% | 1,485 | 775 | -48% | 0 | 0 | — |
case-12 | pass→pass | 8,427 | 1,956 | -77% | 1 | 1 | 0% | 1,299 | 816 | -37% | 0 | 0 | — |
case-13 | fail→pass | 12,016 | 3,418 | -72% | 1 | 1 | 0% | 1,844 | 1,099 | -40% | 0 | 0 | — |
case-14 | pass→pass | 9,682 | 3,503 | -64% | 1 | 1 | 0% | 1,490 | 1,063 | -29% | 0 | 0 | — |
case-15 | fail→pass | 11,091 | 3,780 | -66% | 1 | 1 | 0% | 1,607 | 1,111 | -31% | 0 | 0 | — |
case-16 | pass→pass | 5,708 | 1,954 | -66% | 1 | 1 | 0% | 888 | 789 | -11% | 0 | 0 | — |
case-17 | fail→pass | 4,987 | 3,137 | -37% | 1 | 1 | 0% | 733 | 1,015 | +38% | 0 | 0 | — |
case-18 | pass→pass | 12,578 | 4,777 | -62% | 1 | 1 | 0% | 1,999 | 1,299 | -35% | 0 | 0 | — |
case-19 | pass→pass | 13,797 | 2,147 | -84% | 1 | 1 | 0% | 2,151 | 855 | -60% | 0 | 0 | — |
case-20 | pass→pass | 5,597 | 9,593 | +71% | 1 | 1 | 0% | 990 | 1,653 | +67% | 0 | 0 | — |
case-21 | fail→pass | 9,938 | 5,442 | -45% | 1 | 1 | 0% | 1,626 | 1,505 | -7% | 0 | 0 | — |
case-22 | pass→pass | 5,768 | 2,910 | -50% | 1 | 1 | 0% | 998 | 946 | -5% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted. The headline lift of +55 percentage points is the difference between those two pass rates over the 22 comparable cases.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.