Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Review code changes against accepted ADRs for compliance violations
.claude/skills/ruvnet-adr-review/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-11 | ✗→✓ | ▲ Improved | -18% | 0% |
| case-12 | ✗→✓ | ▲ Improved | 11% | 0% |
| case-13 | ✗→✓ | ▲ Improved | -36% | 0% |
| case-20 | ✗→✓ | ▲ Improved | -56% | 0% |
| case-22 | ✗→✓ | ▲ Improved | -33% | 0% |
Review code changes against accepted Architecture Decision Records to detect violations, drift, and non-compliance.
Before merging a PR, after significant code changes, or as part of a periodic compliance check. Use this to ensure code changes respect the architectural decisions the team has agreed upon.
git diff main...HEAD --name-only (or the specified branch) to list changed files. Then run git diff main...HEAD to get the full diff content.Grep the file for ADR references (ADR-\d+)Grep docs/adr/ for ADRs that mention the changed file paths or modulesmcp__plugin_ruflo-core_ruflo__memory_search with the file path and change summary to find semantically related ADRsRead each relevant ADR file. Focus on:mcp__plugin_ruflo-core_ruflo__agentdb_causal-query to check if any referenced ADRs have been superseded. If so, flag that the code references an outdated decision. ## ADR Compliance Report
### Violations
### Warnings
### Compliant
### Unlinked Changes
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 4,850 | 5,216 | +8% | 1 | 1 | 0% | 884 | 852 | -4% | 0 | 0 | — |
case-06 | pass→fail | 3,137 | 6,636 | +112% | 1 | 1 | 0% | 473 | 991 | +110% | 0 | 0 | — |
case-02 | fail→fail | 7,413 | 4,752 | -36% | 1 | 1 | 0% | 1,202 | 712 | -41% | 0 | 0 | — |
case-03 | fail→fail | 9,296 | 4,684 | -50% | 1 | 1 | 0% | 1,778 | 714 | -60% | 0 | 0 | — |
case-04 | pass→fail | 11,885 | 4,513 | -62% | 1 | 1 | 0% | 2,061 | 832 | -60% | 0 | 0 | — |
case-05 | fail→fail | 9,784 | 8,160 | -17% | 1 | 1 | 0% | 1,121 | 816 | -27% | 0 | 0 | — |
case-07 | fail→fail | 4,803 | 7,376 | +54% | 1 | 1 | 0% | 687 | 826 | +20% | 0 | 0 | — |
case-08 | pass→pass | 9,430 | 5,673 | -40% | 1 | 1 | 0% | 1,727 | 1,509 | -13% | 0 | 0 | — |
case-09 | fail→fail | 8,336 | 4,813 | -42% | 1 | 1 | 0% | 1,469 | 1,554 | +6% | 0 | 0 | — |
case-10 | pass→fail | 9,995 | 6,887 | -31% | 1 | 1 | 0% | 1,770 | 992 | -44% | 0 | 0 | — |
case-11 | fail→pass | 6,511 | 2,219 | -66% | 1 | 1 | 0% | 1,203 | 983 | -18% | 0 | 0 | — |
case-12 | fail→pass | 3,668 | 1,507 | -59% | 1 | 1 | 0% | 737 | 816 | +11% | 0 | 0 | — |
case-13 | fail→pass | 9,007 | 3,072 | -66% | 1 | 1 | 0% | 1,491 | 955 | -36% | 0 | 0 | — |
case-14 | pass→pass | 10,696 | 6,103 | -43% | 1 | 1 | 0% | 2,050 | 1,662 | -19% | 0 | 0 | — |
case-15 | pass→pass | 9,203 | 3,961 | -57% | 1 | 1 | 0% | 1,653 | 1,223 | -26% | 0 | 0 | — |
case-16 | pass→pass | 11,432 | 3,131 | -73% | 1 | 1 | 0% | 1,848 | 1,102 | -40% | 0 | 0 | — |
case-17 | pass→pass | 10,636 | 4,587 | -57% | 1 | 1 | 0% | 1,758 | 1,354 | -23% | 0 | 0 | — |
case-18 | fail→fail | 6,470 | 3,884 | -40% | 1 | 1 | 0% | 1,198 | 1,273 | +6% | 0 | 0 | — |
case-19 | pass→pass | 11,053 | 5,947 | -46% | 1 | 1 | 0% | 1,785 | 1,610 | -10% | 0 | 0 | — |
case-20 | fail→pass | 19,682 | 5,470 | -72% | 1 | 1 | 0% | 3,568 | 1,578 | -56% | 0 | 0 | — |
case-21 | pass→pass | 5,006 | 3,780 | -24% | 1 | 1 | 0% | 934 | 1,162 | +24% | 0 | 0 | — |
case-22 | fail→pass | 7,028 | 2,049 | -71% | 1 | 1 | 0% | 1,318 | 886 | -33% | 0 | 0 | — |
case-23 | fail→fail | 7,652 | 2,037 | -73% | 1 | 1 | 0% | 1,300 | 959 | -26% | 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. 23 cases were attempted, and 16 counted toward the lift figure. The other 7 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. The headline lift of +9 percentage points is the difference between those two pass rates over the 16 comparable cases. 4 cases got worse with the skill loaded, and they are included in that figure.
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.