Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →One-command drift detection. Composes audit-list + oia-audit + audit-trend into a single primitive — finds the most recent audit in `metaharness-audit` namespace, runs a fresh audit against the current repo, diffs them via ADR-152 §3.1 similarity, and alerts when structural distance crosses `--threshold`. Iter 53 of ADR-150 deep integration.
.claude/skills/ruvnet-harness-drift-from-history/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-03 | ✗→✓ | ▲ Improved | 8% | 0% |
| case-05 | ✗→✓ | ▲ Improved | -2% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 23% | 0% |
| case-10 | ✗→✓ | ▲ Improved | -46% | 0% |
| case-12 | ✗→✓ | ▲ Improved | -54% | 0% |
The natural ops question after running oia-audit weekly is "did anything drift?" Before this skill, the answer required a three-step sequence:
bashnpx ruflo metaharness audit-list --format json # → pick a key by hand npx ruflo metaharness oia-audit --format json > /tmp/curr.json npx ruflo metaharness audit-trend \ --baseline-key <picked-key> --current /tmp/curr.json \ --alert-on-distance-below 0.95
This skill collapses it into one command:
bashnpx ruflo metaharness drift-from-history --threshold 0.95
metaharness-audit namespace via audit-list.mjsstartedAt (or --baseline-since 7d skips anything newer than 7 days)oia-audit against the current path--alert-on-distance-below ${threshold}| Constraint | How drift-from-history satisfies it | |---|---| | Removable | Pure subprocess composition over existing scripts — no new @metaharness/* import | | Optional | If oia-audit reports degraded:true, this skill exits 3 with a degraded payload | | Graceful | Empty audit history → exit 2 with hint to seed it; never crashes | | CI-gate | Smoke step 17z16 anchors the dispatcher entry + subcommand listing |
bash$ npx ruflo metaharness drift-from-history --threshold 0.95 # drift-from-history Baseline: audit-2026-06-16T22-58-47-840Z Current: 2026-06-16T23:05:02.231Z Structural similarity: 1 (near-identical) Distance: 0 ✓ similarity ≥ 0.95 — OK
scripts/drift-from-history.mjs
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 8,689 | 4,856 | -44% | 1 | 1 | 0% | 1,070 | 843 | -21% | 0 | 0 | — |
case-02 | fail→fail | 12,287 | 4,138 | -66% | 1 | 1 | 0% | 2,534 | 838 | -67% | 0 | 0 | — |
case-03 | fail→pass | 5,791 | 3,018 | -48% | 1 | 1 | 0% | 698 | 757 | +8% | 0 | 0 | — |
case-04 | pass→pass | 5,819 | 2,469 | -58% | 1 | 1 | 0% | 1,152 | 1,042 | -10% | 0 | 0 | — |
case-05 | fail→pass | 5,939 | 1,920 | -68% | 1 | 1 | 0% | 979 | 957 | -2% | 0 | 0 | — |
case-06 | pass→pass | 3,848 | 1,447 | -62% | 1 | 1 | 0% | 586 | 810 | +38% | 0 | 0 | — |
case-07 | fail→pass | 12,705 | 2,085 | -84% | 1 | 1 | 0% | 641 | 787 | +23% | 0 | 0 | — |
case-08 | pass→pass | 7,931 | 2,011 | -75% | 1 | 1 | 0% | 1,379 | 914 | -34% | 0 | 0 | — |
case-09 | pass→pass | 13,977 | 2,095 | -85% | 1 | 1 | 0% | 1,357 | 984 | -27% | 0 | 0 | — |
case-10 | fail→pass | 12,301 | 3,085 | -75% | 1 | 1 | 0% | 2,009 | 1,077 | -46% | 0 | 0 | — |
case-11 | pass→pass | 8,844 | 2,444 | -72% | 1 | 1 | 0% | 1,840 | 1,096 | -40% | 0 | 0 | — |
case-12 | fail→pass | 9,256 | 5,794 | -37% | 1 | 1 | 0% | 1,652 | 761 | -54% | 0 | 0 | — |
case-13 | fail→pass | 12,232 | 4,687 | -62% | 1 | 1 | 0% | 2,057 | 856 | -58% | 0 | 0 | — |
case-14 | fail→pass | 4,638 | 1,187 | -74% | 1 | 1 | 0% | 867 | 759 | -12% | 0 | 0 | — |
case-15 | fail→pass | 10,950 | 3,779 | -65% | 1 | 1 | 0% | 1,800 | 1,379 | -23% | 0 | 0 | — |
case-16 | fail→pass | 13,799 | 4,963 | -64% | 1 | 1 | 0% | 2,589 | 1,554 | -40% | 0 | 0 | — |
case-17 | fail→pass | 9,915 | 2,188 | -78% | 1 | 1 | 0% | 1,982 | 1,083 | -45% | 0 | 0 | — |
case-18 | fail→pass | 7,819 | 2,544 | -67% | 1 | 1 | 0% | 1,536 | 1,069 | -30% | 0 | 0 | — |
case-19 | fail→pass | 7,962 | 2,133 | -73% | 1 | 1 | 0% | 542 | 996 | +84% | 0 | 0 | — |
case-20 | fail→pass | 6,584 | 3,757 | -43% | 1 | 1 | 0% | 1,280 | 1,311 | +2% | 0 | 0 | — |
case-21 | fail→pass | 4,822 | 1,778 | -63% | 1 | 1 | 0% | 935 | 940 | +1% | 0 | 0 | — |
case-22 | fail→pass | 6,819 | 2,642 | -61% | 1 | 1 | 0% | 1,276 | 1,149 | -10% | 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, and 18 counted toward the lift figure. The other 4 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 +68 percentage points is the difference between those two pass rates over the 18 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.