Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Test authoring guidance
.claude/skills/bilal140202-code-review-testing/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-03 | ✗→✓ | ▲ Improved | -40% | 0% |
| case-10 | ✗→✓ | ▲ Improved | -77% | 0% |
| case-01 | ✗→✓ | ▲ Improved | -57% | 0% |
| case-02 | ✗→✓ | ▲ Improved | -59% | 0% |
| case-07 | ✗→✓ | ▲ Improved | -65% | 0% |
For agent changes prefer integration tests over unit tests. Integration tests are under core/suite and use test_codex to set up a test instance of codex.
Features that change the agent logic MUST add an integration test:
If unit tests are needed, put them in a dedicated test file (_tests.rs). Avoid test-only functions in the main implementation.
Check whether there are existing helpers to make tests more streamlined and readable.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-03 | fail→pass | 11,560 | 5,526 | -52% | 1 | 1 | 0% | 1,730 | 1,046 | -40% | 0 | 0 | — |
case-04 | pass→pass | 13,578 | 14,091 | +4% | 1 | 1 | 0% | 2,441 | 2,451 | +0% | 0 | 0 | — |
case-10 | fail→pass | 13,664 | 3,099 | -77% | 1 | 1 | 0% | 2,099 | 487 | -77% | 0 | 0 | — |
case-01 | fail→pass | 17,374 | 6,039 | -65% | 1 | 1 | 0% | 2,592 | 1,119 | -57% | 0 | 0 | — |
case-02 | fail→pass | 16,286 | 6,310 | -61% | 1 | 1 | 0% | 2,606 | 1,058 | -59% | 0 | 0 | — |
case-05 | pass→pass | 13,223 | 9,633 | -27% | 1 | 1 | 0% | 2,438 | 1,805 | -26% | 0 | 0 | — |
case-06 | pass→pass | 11,435 | 7,902 | -31% | 1 | 1 | 0% | 1,933 | 1,595 | -17% | 0 | 0 | — |
case-07 | fail→pass | 16,903 | 6,148 | -64% | 1 | 1 | 0% | 3,232 | 1,140 | -65% | 0 | 0 | — |
case-08 | fail→pass | 11,271 | 5,347 | -53% | 1 | 1 | 0% | 2,080 | 1,057 | -49% | 0 | 0 | — |
case-09 | fail→pass | 12,353 | 4,485 | -64% | 1 | 1 | 0% | 2,009 | 897 | -55% | 0 | 0 | — |
case-11 | fail→pass | 12,537 | 5,145 | -59% | 1 | 1 | 0% | 2,028 | 991 | -51% | 0 | 0 | — |
case-12 | fail→pass | 15,800 | 4,224 | -73% | 1 | 1 | 0% | 1,799 | 810 | -55% | 0 | 0 | — |
case-13 | fail→pass | 18,353 | 3,637 | -80% | 1 | 1 | 0% | 2,873 | 652 | -77% | 0 | 0 | — |
case-14 | pass→pass | 11,047 | 5,216 | -53% | 1 | 1 | 0% | 2,039 | 1,047 | -49% | 0 | 0 | — |
case-15 | fail→pass | 16,822 | 3,681 | -78% | 1 | 1 | 0% | 2,982 | 745 | -75% | 0 | 0 | — |
case-16 | pass→pass | 14,286 | 4,838 | -66% | 1 | 1 | 0% | 2,059 | 794 | -61% | 0 | 0 | — |
case-17 | fail→pass | 15,629 | 4,433 | -72% | 1 | 1 | 0% | 2,725 | 911 | -67% | 0 | 0 | — |
case-18 | fail→pass | 9,486 | 5,499 | -42% | 1 | 1 | 0% | 1,653 | 1,063 | -36% | 0 | 0 | — |
case-19 | fail→pass | 12,260 | 4,273 | -65% | 1 | 1 | 0% | 1,997 | 753 | -62% | 0 | 0 | — |
case-20 | fail→pass | 14,270 | 4,443 | -69% | 1 | 1 | 0% | 2,312 | 939 | -59% | 0 | 0 | — |
case-21 | pass→pass | 14,937 | 3,761 | -75% | 1 | 1 | 0% | 2,261 | 775 | -66% | 0 | 0 | — |
case-22 | fail→pass | 10,889 | 3,237 | -70% | 1 | 1 | 0% | 1,741 | 705 | -60% | 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 +73 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.