Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Plan a team offsite that earns its cost — the purpose split (connection vs. decisions vs. planning, weighted on purpose), the agenda that alternates work and air, the logistics runbook, and the follow-through that makes Monday different from before. Use when asked plan our team offsite, design two days for the team, make this offsite not a waste, or what do we actually do at the offsite. Produces the purpose weighting, the day designs, the logistics checklist, and the commitments-capture that su
.claude/skills/mohitagw15856-offsite-planner/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-07 | ✗→✓ | ▲ Improved | 46% | 0% |
| case-08 | ✗→✓ | ▲ Improved | 45% | 0% |
| case-11 | ✗→✓ | ▲ Improved | 32% | 0% |
| case-13 | ✗→✓ | ▲ Improved | 31% | 0% |
| case-17 | ✗→✓ | ▲ Improved | 28% | 0% |
Offsites fail at the extremes: the all-work version (eight hours of conference room that could have been at the office, plus travel) and the all-vibes version (fun had, nothing changed). The honest design starts by weighting the purposes — connection, decisions, planning — because they want different formats, and pretending one agenda serves all three equally serves none. Then: sessions designed like workshops (artifacts, not discussions — workshop-designer rules), air between them (the corridor conversations are half the value; schedule the corridors), and the follow-through discipline, because an offsite's ROI is measured two weeks later, on Mondays.
Ask for these if not provided:
The split + the why from the team's current truth · what success means under it]
| Block | Purpose served | Format → artifact | Air after | |---|---|---|---| Energy arc: heavy work early-day, decisions pre-fatigue, connection where it belongs]
Venue/room fit · travel · food · the hybrid rig + test · owner: (name)]
Last session: commitments (owner+date) + the alignment note · T+2 weeks: the check-in on movement, calendared now]
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 33,710 | 21,311 | -37% | 1 | 1 | 0% | 6,240 | 5,108 | -18% | 0 | 0 | — |
case-02 | fail→fail | 22,835 | 22,120 | -3% | 1 | 1 | 0% | 4,331 | 5,074 | +17% | 0 | 0 | — |
case-03 | pass→fail | 30,561 | 20,753 | -32% | 1 | 1 | 0% | 5,696 | 5,107 | -10% | 0 | 0 | — |
case-04 | pass→pass | 10,709 | 15,099 | +41% | 1 | 1 | 0% | 2,036 | 3,880 | +91% | 0 | 0 | — |
case-05 | pass→pass | 15,090 | 15,104 | +0% | 1 | 1 | 0% | 2,825 | 3,700 | +31% | 0 | 0 | — |
case-06 | pass→pass | 11,669 | 11,564 | -1% | 1 | 1 | 0% | 2,058 | 3,131 | +52% | 0 | 0 | — |
case-07 | fail→pass | 17,326 | 17,214 | -1% | 1 | 1 | 0% | 2,799 | 4,082 | +46% | 0 | 0 | — |
case-08 | fail→pass | 19,782 | 17,844 | -10% | 1 | 1 | 0% | 2,957 | 4,291 | +45% | 0 | 0 | — |
case-09 | fail→fail | 18,438 | 19,369 | +5% | 1 | 1 | 0% | 2,996 | 4,054 | +35% | 0 | 0 | — |
case-10 | fail→fail | 19,235 | 14,320 | -26% | 1 | 1 | 0% | 3,468 | 3,713 | +7% | 0 | 0 | — |
case-11 | fail→pass | 19,164 | 17,970 | -6% | 1 | 1 | 0% | 3,172 | 4,192 | +32% | 0 | 0 | — |
case-12 | fail→fail | 18,842 | 17,350 | -8% | 1 | 1 | 0% | 3,065 | 4,007 | +31% | 0 | 0 | — |
case-13 | fail→pass | 16,812 | 14,033 | -17% | 1 | 1 | 0% | 2,637 | 3,462 | +31% | 0 | 0 | — |
case-14 | fail→fail | 19,653 | 18,869 | -4% | 1 | 1 | 0% | 3,275 | 4,258 | +30% | 0 | 0 | — |
case-15 | fail→fail | 14,354 | 17,070 | +19% | 1 | 1 | 0% | 2,425 | 3,817 | +57% | 0 | 0 | — |
case-16 | fail→fail | 14,936 | 17,551 | +18% | 1 | 1 | 0% | 2,561 | 4,056 | +58% | 0 | 0 | — |
case-17 | fail→pass | 18,777 | 16,155 | -14% | 1 | 1 | 0% | 2,922 | 3,735 | +28% | 0 | 0 | — |
case-18 | fail→fail | 16,680 | 15,118 | -9% | 1 | 1 | 0% | 2,650 | 3,526 | +33% | 0 | 0 | — |
case-19 | pass→pass | 15,488 | 18,822 | +22% | 1 | 1 | 0% | 2,596 | 4,178 | +61% | 0 | 0 | — |
case-20 | fail→pass | 18,485 | 20,456 | +11% | 1 | 1 | 0% | 3,000 | 4,489 | +50% | 0 | 0 | — |
case-21 | fail→fail | 14,941 | 16,736 | +12% | 1 | 1 | 0% | 2,381 | 3,859 | +62% | 0 | 0 | — |
case-22 | fail→fail | 18,743 | 16,202 | -14% | 1 | 1 | 0% | 3,105 | 3,806 | +23% | 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 +23 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.