Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →AI DevKit · Review and improve documentation for novice users. Use when users ask to review docs, improve documentation, audit README files, evaluate API docs, review guides, or improve technical writing.
.claude/skills/bilal140202-technical-writer/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | -6% | 0% |
| case-05 | ✗→✓ | ▲ Improved | -20% | 0% |
| case-08 | ✗→✓ | ▲ Improved | 53% | 0% |
| case-19 | ✓→✗ | ▼ Worse | -3% | 0% |
| case-04 | ✓→✓ | ＝ Same ✓ | 5% | 0% |
Review documentation as a novice would experience it. Suggest concrete improvements.
| Rationalization | Why It's Wrong | Do Instead | |---|---|---| | "Developers will figure it out" | Novice users won't | Write for the least experienced reader | | "The code example speaks for itself" | Examples without context confuse | Add what it does and when to use it | | "Too much detail clutters the doc" | Missing detail blocks users | Include prerequisites and edge cases |
## [Document Name]
| Aspect | Rating | Notes |
|--------|--------|-------|
| Clarity | X/5 | ... |
| Completeness | X/5 | ... |
| Actionability | X/5 | ... |
| Structure | X/5 | ... |
**Issues:**
1. [High] Description (line X)
2. [Medium] Description (line X)
**Suggested Fixes:**
- Concrete fix with example text| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-02 | fail→fail | 8,033 | 2,494 | -69% | 1 | 1 | 0% | 1,230 | 767 | -38% | 0 | 0 | — |
case-03 | fail→fail | 2,084 | 3,490 | +67% | 1 | 1 | 0% | 297 | 935 | +215% | 0 | 0 | — |
case-01 | fail→pass | 6,124 | 3,657 | -40% | 1 | 1 | 0% | 1,067 | 1,006 | -6% | 0 | 0 | — |
case-04 | pass→pass | 8,533 | 6,303 | -26% | 1 | 1 | 0% | 1,568 | 1,641 | +5% | 0 | 0 | — |
case-05 | fail→pass | 12,605 | 7,768 | -38% | 1 | 1 | 0% | 2,138 | 1,709 | -20% | 0 | 0 | — |
case-06 | pass→pass | 13,596 | 6,974 | -49% | 1 | 1 | 0% | 2,359 | 1,676 | -29% | 0 | 0 | — |
case-07 | pass→pass | 8,509 | 4,253 | -50% | 1 | 1 | 0% | 1,318 | 1,070 | -19% | 0 | 0 | — |
case-08 | fail→pass | 5,729 | 5,672 | -1% | 1 | 1 | 0% | 919 | 1,402 | +53% | 0 | 0 | — |
case-09 | pass→pass | 6,656 | 3,432 | -48% | 1 | 1 | 0% | 1,130 | 983 | -13% | 0 | 0 | — |
case-10 | pass→pass | 9,653 | 7,158 | -26% | 1 | 1 | 0% | 1,551 | 1,756 | +13% | 0 | 0 | — |
case-11 | pass→pass | 10,370 | 6,324 | -39% | 1 | 1 | 0% | 1,849 | 1,590 | -14% | 0 | 0 | — |
case-12 | pass→pass | 10,340 | 6,043 | -42% | 1 | 1 | 0% | 1,574 | 1,385 | -12% | 0 | 0 | — |
case-13 | pass→pass | 10,766 | 4,712 | -56% | 1 | 1 | 0% | 1,864 | 1,244 | -33% | 0 | 0 | — |
case-14 | pass→pass | 9,515 | 6,735 | -29% | 1 | 1 | 0% | 1,547 | 1,619 | +5% | 0 | 0 | — |
case-15 | pass→pass | 9,527 | 4,833 | -49% | 1 | 1 | 0% | 1,614 | 1,136 | -30% | 0 | 0 | — |
case-21 | pass→pass | 3,777 | 9,987 | +164% | 1 | 1 | 0% | 757 | 2,290 | +203% | 0 | 0 | — |
case-16 | pass→pass | 12,040 | 7,757 | -36% | 1 | 1 | 0% | 1,853 | 1,560 | -16% | 0 | 0 | — |
case-17 | pass→pass | 11,854 | 7,173 | -39% | 1 | 1 | 0% | 1,862 | 1,532 | -18% | 0 | 0 | — |
case-18 | pass→pass | 12,825 | 8,373 | -35% | 1 | 1 | 0% | 2,244 | 1,879 | -16% | 0 | 0 | — |
case-19 | pass→fail | 4,731 | 3,306 | -30% | 1 | 1 | 0% | 820 | 796 | -3% | 0 | 0 | — |
case-20 | pass→pass | 4,088 | 7,633 | +87% | 1 | 1 | 0% | 926 | 1,896 | +105% | 0 | 0 | — |
case-22 | fail→fail | 6,801 | 11,593 | +70% | 1 | 1 | 0% | 1,291 | 2,628 | +104% | 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 +9 percentage points is the difference between those two pass rates over the 22 comparable cases. 1 case got worse with the skill loaded, and it is 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.