Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Generates or fills gaps in project documentation (README, AGENTS.md, /docs, inline comments).
.claude/skills/tariux-documentation/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 119% | 0% |
| case-02 | ✗→✓ | ▲ Improved | -48% | 0% |
| case-05 | ✗→✓ | ▲ Improved | -12% | 0% |
| case-07 | ✗→✓ | ▲ Improved | -4% | 0% |
| case-08 | ✗→✓ | ▲ Improved | 34% | 0% |
System Instruction: Documentation Mode
You manage project documentation. Use a clean, concise style. Do not add marketing fluff.
If existing documentation is present (README, inline docs, /docs folder):
If no documentation exists:
AGENTS.md should include: project overview, technology stack, folder structure, coding conventions, and key constraints.
Output the file paths and content. Stop immediately after.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 7,394 | 15,696 | +112% | 1 | 1 | 0% | 1,014 | 2,223 | +119% | 0 | 0 | — |
case-02 | fail→pass | 45,841 | 19,396 | -58% | 1 | 1 | 0% | 8,046 | 4,214 | -48% | 0 | 0 | — |
case-03 | pass→pass | 27,747 | 14,629 | -47% | 1 | 1 | 0% | 4,975 | 2,807 | -44% | 0 | 0 | — |
case-04 | pass→pass | 30,514 | 13,173 | -57% | 1 | 1 | 0% | 5,298 | 2,677 | -49% | 0 | 0 | — |
case-05 | fail→pass | 11,789 | 10,328 | -12% | 1 | 1 | 0% | 2,376 | 2,098 | -12% | 0 | 0 | — |
case-06 | pass→pass | 16,150 | 17,010 | +5% | 1 | 1 | 0% | 2,890 | 3,440 | +19% | 0 | 0 | — |
case-07 | fail→pass | 12,008 | 9,613 | -20% | 1 | 1 | 0% | 2,126 | 2,043 | -4% | 0 | 0 | — |
case-08 | fail→pass | 13,045 | 15,290 | +17% | 1 | 1 | 0% | 2,222 | 2,977 | +34% | 0 | 0 | — |
case-09 | pass→pass | 14,483 | 6,921 | -52% | 1 | 1 | 0% | 2,855 | 1,612 | -44% | 0 | 0 | — |
case-10 | pass→pass | 5,022 | 3,476 | -31% | 1 | 1 | 0% | 795 | 820 | +3% | 0 | 0 | — |
case-11 | pass→pass | 23,869 | 13,931 | -42% | 1 | 1 | 0% | 4,189 | 2,834 | -32% | 0 | 0 | — |
case-12 | pass→pass | 25,465 | 15,897 | -38% | 1 | 1 | 0% | 4,315 | 3,121 | -28% | 0 | 0 | — |
case-13 | fail→pass | 46,717 | 20,240 | -57% | 1 | 1 | 0% | 8,211 | 4,025 | -51% | 0 | 0 | — |
case-14 | fail→fail | 9,924 | 4,973 | -50% | 1 | 1 | 0% | 1,884 | 1,167 | -38% | 0 | 0 | — |
case-15 | pass→pass | 12,127 | 11,900 | -2% | 1 | 1 | 0% | 2,347 | 2,590 | +10% | 0 | 0 | — |
case-16 | pass→fail | 10,171 | 10,007 | -2% | 1 | 1 | 0% | 1,902 | 2,280 | +20% | 0 | 0 | — |
case-17 | pass→pass | 3,767 | 2,480 | -34% | 1 | 1 | 0% | 662 | 634 | -4% | 0 | 0 | — |
case-18 | pass→fail | 12,716 | 15,275 | +20% | 1 | 1 | 0% | 1,973 | 2,997 | +52% | 0 | 0 | — |
case-19 | pass→pass | 28,238 | 14,008 | -50% | 1 | 1 | 0% | 4,970 | 2,900 | -42% | 0 | 0 | — |
case-20 | pass→pass | 11,266 | 3,685 | -67% | 1 | 1 | 0% | 1,877 | 869 | -54% | 0 | 0 | — |
case-21 | fail→pass | 18,783 | 17,656 | -6% | 1 | 1 | 0% | 2,950 | 3,590 | +22% | 0 | 0 | — |
case-22 | pass→pass | 44,693 | 15,072 | -66% | 1 | 1 | 0% | 8,210 | 3,004 | -63% | 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 +23 percentage points is the difference between those two pass rates over the 22 comparable cases. 2 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.