Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Identifies outdated elements in provided content and suggests updates to maintain freshness. Finds statistics, dates, and examples that need updating. Use PROACTIVELY for older content.
.claude/skills/dokhacgiakhoa-seo-content-refresher/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-02 | ✗→✓ | ▲ Improved | 10% | 0% |
| case-04 | ✗→✓ | ▲ Improved | 1% | 0% |
| case-18 | ✓→✗ | ▼ Worse | 37% | 0% |
| case-03 | ✓→✓ | ＝ Same ✓ | 15% | 0% |
| case-05 | ✓→✓ | ＝ Same ✓ | -18% | 0% |
resources/implementation-playbook.md.You are a content freshness specialist identifying update opportunities in existing content.
Update Priorities:
High Priority (Immediate):
Medium Priority (This Month):
Content Refresh Plan:
Page: [URL]
Last Updated: [Date]
Priority: High/Medium/Low
Refresh Actions:
- Update statistics from 2023 to 2025
- Add section on [new trend]
- Refresh examples with current ones
- Update meta title with "2025"Deliverables:
Refresh Tactics:
Freshness Signals:
Platform Implementation:
Focus on meaningful updates that add value. Identify specific elements that need refreshing.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 15,507 | 25,420 | +64% | 1 | 1 | 0% | 2,456 | 3,564 | +45% | 0 | 0 | — |
case-02 | fail→pass | 19,947 | 18,709 | -6% | 1 | 1 | 0% | 3,067 | 3,359 | +10% | 0 | 0 | — |
case-03 | pass→pass | 8,966 | 5,739 | -36% | 1 | 1 | 0% | 1,423 | 1,643 | +15% | 0 | 0 | — |
case-04 | fail→pass | 16,822 | 11,256 | -33% | 1 | 1 | 0% | 2,400 | 2,424 | +1% | 0 | 0 | — |
case-05 | pass→pass | 13,573 | 7,104 | -48% | 1 | 1 | 0% | 2,323 | 1,912 | -18% | 0 | 0 | — |
case-06 | pass→pass | 5,970 | 6,810 | +14% | 1 | 1 | 0% | 970 | 1,837 | +89% | 0 | 0 | — |
case-07 | pass→pass | 5,912 | 5,293 | -10% | 1 | 1 | 0% | 1,138 | 1,599 | +41% | 0 | 0 | — |
case-08 | pass→pass | 16,361 | 15,697 | -4% | 1 | 1 | 0% | 3,043 | 3,577 | +18% | 0 | 0 | — |
case-09 | pass→pass | 21,108 | 16,293 | -23% | 1 | 1 | 0% | 3,332 | 3,018 | -9% | 0 | 0 | — |
case-10 | pass→pass | 16,964 | 19,408 | +14% | 1 | 1 | 0% | 2,759 | 3,383 | +23% | 0 | 0 | — |
case-11 | pass→pass | 18,425 | 17,351 | -6% | 1 | 1 | 0% | 2,834 | 3,427 | +21% | 0 | 0 | — |
case-12 | pass→pass | 10,812 | 12,524 | +16% | 1 | 1 | 0% | 1,801 | 2,533 | +41% | 0 | 0 | — |
case-13 | pass→pass | 10,225 | 6,416 | -37% | 1 | 1 | 0% | 1,754 | 1,745 | -1% | 0 | 0 | — |
case-14 | pass→pass | 14,578 | 14,204 | -3% | 1 | 1 | 0% | 2,338 | 2,846 | +22% | 0 | 0 | — |
case-15 | pass→pass | 7,544 | 4,563 | -40% | 1 | 1 | 0% | 1,192 | 1,478 | +24% | 0 | 0 | — |
case-16 | pass→pass | 14,373 | 12,659 | -12% | 1 | 1 | 0% | 2,194 | 2,798 | +28% | 0 | 0 | — |
case-17 | fail→fail | 19,690 | 19,279 | -2% | 1 | 1 | 0% | 3,244 | 3,463 | +7% | 0 | 0 | — |
case-18 | pass→fail | 7,989 | 20,857 | +161% | 1 | 1 | 0% | 1,488 | 2,034 | +37% | 0 | 0 | — |
case-19 | pass→pass | 12,539 | 14,069 | +12% | 1 | 1 | 0% | 2,137 | 2,891 | +35% | 0 | 0 | — |
case-20 | pass→pass | 27,350 | 21,475 | -21% | 1 | 1 | 0% | 5,036 | 3,808 | -24% | 0 | 0 | — |
case-21 | pass→pass | 24,424 | 23,014 | -6% | 1 | 1 | 0% | 4,090 | 4,175 | +2% | 0 | 0 | — |
case-22 | pass→pass | 17,161 | 14,351 | -16% | 1 | 1 | 0% | 2,758 | 3,045 | +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. The headline lift of +5 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.