Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →When the user wants to plan PR, write a press release, or manage media relations. Also use when the user mentions "public relations," "PR," "press release," "media relations," "news release," "journalist," "media coverage," "product announcement," or "earned media." For on-site "As Seen In" or logos page, use press-coverage-page-generator.
.claude/skills/mkurman-public-relations/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | -6% | 0% |
| case-02 | ✗→✓ | ▲ Improved | 26% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 11% | 0% |
| case-04 | ✗→✓ | ▲ Improved | -1% | 0% |
| case-06 | ✗→✓ | ▲ Improved | 12% | 0% |
---|-----------| | Header | Logo; contact (name, title, email, phone); "FOR IMMEDIATE RELEASE" or embargo | | Headline | Under 100 chars; strong action verbs; "Why should I care?" | | Subheadline | Optional; additional context | | Dateline | City, state, date | | Lead | 50–75 words; all 5 W's (Who, What, When, Where, Why) | | Body | 1–2 paragraphs; inverted pyramid; most newsworthy first | | Quote | Executive/stakeholder; perspective, not fact repetition | | Boilerplate | 2–3 sentence company description | | Media contact | Name, email, phone |
Journalist should understand the full story from the lead alone. Specific details, not vague language ("important update" → what changed and impact).
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 12,941 | 10,033 | -22% | 1 | 1 | 0% | 2,089 | 1,966 | -6% | 0 | 0 | — |
case-02 | fail→pass | 12,164 | 12,668 | +4% | 1 | 1 | 0% | 2,054 | 2,582 | +26% | 0 | 0 | — |
case-03 | fail→pass | 13,412 | 11,621 | -13% | 1 | 1 | 0% | 2,184 | 2,426 | +11% | 0 | 0 | — |
case-04 | fail→pass | 12,034 | 9,084 | -25% | 1 | 1 | 0% | 1,913 | 1,893 | -1% | 0 | 0 | — |
case-05 | pass→pass | 11,338 | 10,965 | -3% | 1 | 1 | 0% | 1,742 | 2,338 | +34% | 0 | 0 | — |
case-06 | fail→pass | 11,620 | 10,156 | -13% | 1 | 1 | 0% | 1,929 | 2,164 | +12% | 0 | 0 | — |
case-07 | pass→pass | 11,246 | 11,255 | +0% | 1 | 1 | 0% | 1,817 | 2,424 | +33% | 0 | 0 | — |
case-08 | pass→pass | 15,210 | 8,549 | -44% | 1 | 1 | 0% | 2,439 | 1,702 | -30% | 0 | 0 | — |
case-09 | fail→pass | 12,289 | 10,144 | -17% | 1 | 1 | 0% | 1,900 | 1,960 | +3% | 0 | 0 | — |
case-10 | pass→pass | 10,426 | 10,829 | +4% | 1 | 1 | 0% | 1,697 | 2,084 | +23% | 0 | 0 | — |
case-11 | fail→pass | 11,686 | 12,580 | +8% | 1 | 1 | 0% | 1,864 | 2,455 | +32% | 0 | 0 | — |
case-12 | fail→pass | 10,782 | 8,466 | -21% | 1 | 1 | 0% | 1,689 | 1,730 | +2% | 0 | 0 | — |
case-13 | pass→pass | 11,814 | 9,993 | -15% | 1 | 1 | 0% | 1,897 | 2,115 | +11% | 0 | 0 | — |
case-14 | pass→pass | 12,023 | 11,033 | -8% | 1 | 1 | 0% | 1,925 | 2,202 | +14% | 0 | 0 | — |
case-15 | fail→pass | 11,603 | 10,032 | -14% | 1 | 1 | 0% | 1,853 | 2,105 | +14% | 0 | 0 | — |
case-16 | fail→pass | 10,320 | 10,951 | +6% | 1 | 1 | 0% | 1,710 | 2,297 | +34% | 0 | 0 | — |
case-17 | fail→pass | 12,914 | 10,108 | -22% | 1 | 1 | 0% | 1,955 | 2,136 | +9% | 0 | 0 | — |
case-18 | fail→fail | 11,853 | 10,710 | -10% | 1 | 1 | 0% | 1,856 | 2,253 | +21% | 0 | 0 | — |
case-19 | pass→pass | 12,511 | 11,168 | -11% | 1 | 1 | 0% | 1,803 | 2,311 | +28% | 0 | 0 | — |
case-20 | pass→pass | 11,193 | 10,713 | -4% | 1 | 1 | 0% | 1,812 | 1,952 | +8% | 0 | 0 | — |
case-21 | pass→pass | 14,037 | 14,418 | +3% | 1 | 1 | 0% | 2,143 | 2,517 | +17% | 0 | 0 | — |
case-22 | fail→pass | 14,206 | 13,302 | -6% | 1 | 1 | 0% | 2,107 | 2,345 | +11% | 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 +55 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.