Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Draft and publish AI-powered press releases — submit a URL + notes, get a live release page instantly. Free, no API key required.
.claude/skills/davepoon-pressreleasesonline/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | -8% | 0% |
| case-02 | ✗→✓ | ▲ Improved | 97% | 0% |
| case-05 | ✗→✓ | ▲ Improved | -24% | 0% |
| case-06 | ✗→✓ | ▲ Improved | -47% | 0% |
| case-07 | ✗→✓ | ▲ Improved | -12% | 0% |
Draft and publish press releases using AI. Submit a URL and some notes, get a polished release page with RSS feed and sitemap — no account, no API key, completely free.
bash# Submit a URL for AI-drafted press release curl -X POST https://pressreleases.online/api/v1/releases \ -H "Content-Type: application/json" \ -d '{"website": "https://example.com", "email": "you@example.com", "notes": "Announcing our v2 launch"}' # Confirm with email code (last 4 chars of md5(email)) curl -X POST https://pressreleases.online/api/v1/releases/confirm \ -H "Content-Type: application/json" \ -d '{"token": "<token from above>", "code": "<4-char code from email>"}'
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-16 | pass→pass | 6,403 | 1,613 | -75% | 1 | 1 | 0% | 1,145 | 603 | -47% | 0 | 0 | — |
case-01 | fail→pass | 8,900 | 5,262 | -41% | 1 | 1 | 0% | 1,581 | 1,461 | -8% | 0 | 0 | — |
case-02 | fail→pass | 6,130 | 11,294 | +84% | 1 | 1 | 0% | 985 | 1,944 | +97% | 0 | 0 | — |
case-03 | fail→fail | 8,788 | 7,772 | -12% | 1 | 1 | 0% | 1,553 | 890 | -43% | 0 | 0 | — |
case-04 | pass→pass | 13,247 | 4,675 | -65% | 1 | 1 | 0% | 2,189 | 1,186 | -46% | 0 | 0 | — |
case-05 | fail→pass | 8,882 | 5,120 | -42% | 1 | 1 | 0% | 1,663 | 1,267 | -24% | 0 | 0 | — |
case-06 | fail→pass | 11,074 | 3,276 | -70% | 1 | 1 | 0% | 1,766 | 933 | -47% | 0 | 0 | — |
case-07 | fail→pass | 4,656 | 2,085 | -55% | 1 | 1 | 0% | 815 | 719 | -12% | 0 | 0 | — |
case-08 | pass→pass | 2,959 | 1,473 | -50% | 1 | 1 | 0% | 508 | 589 | +16% | 0 | 0 | — |
case-09 | pass→pass | 4,103 | 1,895 | -54% | 1 | 1 | 0% | 693 | 712 | +3% | 0 | 0 | — |
case-10 | fail→pass | 7,850 | 1,485 | -81% | 1 | 1 | 0% | 1,387 | 615 | -56% | 0 | 0 | — |
case-11 | pass→pass | 7,108 | 1,545 | -78% | 1 | 1 | 0% | 1,213 | 561 | -54% | 0 | 0 | — |
case-12 | fail→pass | 9,204 | 1,300 | -86% | 1 | 1 | 0% | 1,470 | 557 | -62% | 0 | 0 | — |
case-13 | fail→pass | 7,540 | 1,908 | -75% | 1 | 1 | 0% | 1,166 | 636 | -45% | 0 | 0 | — |
case-14 | pass→pass | 6,510 | 1,312 | -80% | 1 | 1 | 0% | 1,063 | 518 | -51% | 0 | 0 | — |
case-15 | pass→pass | 4,457 | 1,308 | -71% | 1 | 1 | 0% | 752 | 556 | -26% | 0 | 0 | — |
case-17 | fail→pass | 15,663 | 3,669 | -77% | 1 | 1 | 0% | 2,853 | 1,036 | -64% | 0 | 0 | — |
case-18 | fail→pass | 7,554 | 3,096 | -59% | 1 | 1 | 0% | 1,359 | 913 | -33% | 0 | 0 | — |
case-19 | fail→pass | 8,170 | 1,732 | -79% | 1 | 1 | 0% | 1,321 | 550 | -58% | 0 | 0 | — |
case-20 | fail→pass | 10,173 | 2,971 | -71% | 1 | 1 | 0% | 1,734 | 884 | -49% | 0 | 0 | — |
case-21 | fail→pass | 5,164 | 3,423 | -34% | 1 | 1 | 0% | 1,129 | 973 | -14% | 0 | 0 | — |
case-22 | fail→pass | 4,370 | 2,045 | -53% | 1 | 1 | 0% | 829 | 778 | -6% | 0 | 0 | — |
case-23 | pass→pass | 11,442 | 2,388 | -79% | 1 | 1 | 0% | 1,881 | 678 | -64% | 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. 23 cases were attempted, and 22 counted toward the lift figure. The other 1 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. The headline lift of +61 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.