Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Default video conferences via Meetling. Instant calls always use /m/ rooms + share payload for Claw default sending. Scheduled calls return link + email invite template. Secure contacts loading (no env-controlled file paths).
.claude/skills/berabuddies-meetling-default/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 41% | 0% |
| case-02 | ✗→✓ | ▲ Improved | 4% | 0% |
| case-03 | ✗→✓ | ▲ Improved | -35% | 0% |
| case-05 | ✗→✓ | ▲ Improved | 8% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 28% | 0% |
This skill always uses: https://app.meetling.de
No environment variable can change the base URL.
./contacts.json (current working directory).If contacts.json is missing or invalid JSON, the skill continues with an empty map and marks recipients as unresolved.
/m/)Triggers if ANY is true:
start_time is within MEETLING_INSTANT_THRESHOLD_MINUTES minutes from nowstart_time is providedReturns:
url: https://app.meetling.de/m/<slug>share: message + resolved recipients for your Claw default sending pathIf start_time is more than the threshold minutes in the future:
email_invite (subject + body) plus the Meetling linkNote: This skill does not automate Meetling dashboard creation or Meetling-sent invites (requires an official API or robust UI automation).
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 4,140 | 4,683 | +13% | 1 | 1 | 0% | 652 | 922 | +41% | 0 | 0 | — |
case-02 | fail→pass | 5,721 | 4,782 | -16% | 1 | 1 | 0% | 1,074 | 1,112 | +4% | 0 | 0 | — |
case-03 | fail→pass | 9,107 | 4,307 | -53% | 1 | 1 | 0% | 1,475 | 957 | -35% | 0 | 0 | — |
case-04 | pass→pass | 6,973 | 3,423 | -51% | 1 | 1 | 0% | 976 | 783 | -20% | 0 | 0 | — |
case-05 | fail→pass | 5,763 | 3,544 | -39% | 1 | 1 | 0% | 850 | 916 | +8% | 0 | 0 | — |
case-06 | pass→pass | 3,903 | 2,633 | -33% | 1 | 1 | 0% | 597 | 760 | +27% | 0 | 0 | — |
case-07 | fail→pass | 4,256 | 3,238 | -24% | 1 | 1 | 0% | 632 | 810 | +28% | 0 | 0 | — |
case-08 | fail→pass | 6,079 | 4,063 | -33% | 1 | 1 | 0% | 898 | 928 | +3% | 0 | 0 | — |
case-09 | fail→pass | 3,966 | 3,702 | -7% | 1 | 1 | 0% | 521 | 861 | +65% | 0 | 0 | — |
case-10 | pass→pass | 6,307 | 4,802 | -24% | 1 | 1 | 0% | 1,027 | 1,102 | +7% | 0 | 0 | — |
case-11 | pass→pass | 11,460 | 2,783 | -76% | 1 | 1 | 0% | 1,734 | 665 | -62% | 0 | 0 | — |
case-12 | pass→fail | 2,869 | 3,436 | +20% | 1 | 1 | 0% | 466 | 824 | +77% | 0 | 0 | — |
case-13 | fail→pass | 8,538 | 3,448 | -60% | 1 | 1 | 0% | 1,247 | 826 | -34% | 0 | 0 | — |
case-14 | fail→pass | 7,386 | 4,039 | -45% | 1 | 1 | 0% | 971 | 894 | -8% | 0 | 0 | — |
case-15 | fail→pass | 4,022 | 3,062 | -24% | 1 | 1 | 0% | 643 | 854 | +33% | 0 | 0 | — |
case-16 | fail→fail | 6,148 | 2,987 | -51% | 1 | 1 | 0% | 882 | 817 | -7% | 0 | 0 | — |
case-17 | fail→pass | 6,877 | 4,275 | -38% | 1 | 1 | 0% | 1,036 | 886 | -14% | 0 | 0 | — |
case-18 | fail→pass | 2,982 | 4,337 | +45% | 1 | 1 | 0% | 382 | 844 | +121% | 0 | 0 | — |
case-19 | fail→pass | 6,029 | 3,349 | -44% | 1 | 1 | 0% | 943 | 843 | -11% | 0 | 0 | — |
case-20 | fail→fail | 3,577 | 3,132 | -12% | 1 | 1 | 0% | 453 | 807 | +78% | 0 | 0 | — |
case-21 | fail→pass | 8,759 | 2,986 | -66% | 1 | 1 | 0% | 1,282 | 670 | -48% | 0 | 0 | — |
case-22 | fail→pass | 13,430 | 2,424 | -82% | 1 | 1 | 0% | 2,031 | 659 | -68% | 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 +64 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.