Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Spin up a tabletop RPG one-shot or a session-zero for a new campaign — a hook, a map of the first adventure, NPCs, and encounters tuned to your party. Use when asked to start a D&D campaign, run a one-shot, help me DM, session zero, or make me an adventure for my party. Produces a premise and hook, a session-zero framework (tone, safety tools, expectations), a first-adventure outline with beats and branches, ready-to-run NPCs and encounters scaled to party level/size, and improv fallbacks for wh
.claude/skills/mohitagw15856-dnd-campaign-starter/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-02 | ✗→✓ | ▲ Improved | -31% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 7% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 26% | 0% |
| case-08 | ✗→✓ | ▲ Improved | 67% | 0% |
| case-09 | ✗→✓ | ▲ Improved | 298% | 0% |
The blank page is the hardest part of running a game. This gives a new or busy DM a runnable start: a hook players will bite on, a session-zero to set tone and boundaries, and a first adventure with enough structure to run confidently — plus the flexibility for when the party inevitably does something you didn't plan.
Ask for these if not provided:
Premise: one-line pitch]. Hook: why the party acts, now].
Session zero: tone x] · boundaries include/avoid] · safety tools e.g. X-card] · expectations scheduling, PvP, etc.].
First adventure
NPCs: name — want — voice] ×3. Encounters: encounter — scaled to level/size — the twist] ×2–3.
If players go off-script: improv fallbacks + a rumor/name table].
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 72,294 | 46,630 | -35% | 1 | 1 | 0% | 7,426 | 7,100 | -4% | 0 | 0 | — |
case-02 | fail→pass | 54,541 | 31,135 | -43% | 1 | 1 | 0% | 6,837 | 4,690 | -31% | 0 | 0 | — |
case-03 | fail→pass | 60,206 | 54,365 | -10% | 1 | 1 | 0% | 4,608 | 4,944 | +7% | 0 | 0 | — |
case-04 | fail→fail | 33,420 | 29,090 | -13% | 1 | 1 | 0% | 4,398 | 4,266 | -3% | 0 | 0 | — |
case-05 | fail→fail | 21,582 | 35,576 | +65% | 1 | 1 | 0% | 1,987 | 4,851 | +144% | 0 | 0 | — |
case-06 | pass→pass | 28,338 | 31,747 | +12% | 1 | 1 | 0% | 2,853 | 4,264 | +49% | 0 | 0 | — |
case-07 | fail→pass | 26,162 | 25,716 | -2% | 1 | 1 | 0% | 3,617 | 4,547 | +26% | 0 | 0 | — |
case-08 | fail→pass | 16,965 | 27,335 | +61% | 1 | 1 | 0% | 2,653 | 4,435 | +67% | 0 | 0 | — |
case-09 | fail→pass | 9,728 | 22,535 | +132% | 1 | 1 | 0% | 923 | 3,671 | +298% | 0 | 0 | — |
case-10 | fail→pass | 20,222 | 29,317 | +45% | 1 | 1 | 0% | 3,233 | 4,169 | +29% | 0 | 0 | — |
case-11 | fail→fail | 82,922 | 34,643 | -58% | 1 | 1 | 0% | 2,949 | 4,589 | +56% | 0 | 0 | — |
case-12 | fail→fail | 47,636 | 34,022 | -29% | 1 | 1 | 0% | 3,014 | 4,506 | +50% | 0 | 0 | — |
case-13 | pass→pass | 32,183 | 28,345 | -12% | 1 | 1 | 0% | 3,232 | 4,142 | +28% | 0 | 0 | — |
case-14 | pass→pass | 28,054 | 37,354 | +33% | 1 | 1 | 0% | 3,043 | 4,855 | +60% | 0 | 0 | — |
case-15 | fail→fail | 43,797 | 28,621 | -35% | 1 | 1 | 0% | 2,814 | 3,700 | +31% | 0 | 0 | — |
case-16 | pass→pass | 32,801 | 26,199 | -20% | 1 | 1 | 0% | 3,316 | 4,225 | +27% | 0 | 0 | — |
case-17 | pass→pass | 16,794 | 107,106 | +538% | 1 | 1 | 0% | 1,586 | 1,779 | +12% | 0 | 0 | — |
case-18 | pass→pass | 25,793 | 24,717 | -4% | 1 | 1 | 0% | 3,636 | 4,183 | +15% | 0 | 0 | — |
case-19 | pass→pass | 35,767 | 27,332 | -24% | 1 | 1 | 0% | 3,117 | 4,377 | +40% | 0 | 0 | — |
case-20 | pass→pass | 23,588 | 26,117 | +11% | 1 | 1 | 0% | 3,380 | 4,847 | +43% | 0 | 0 | — |
case-21 | pass→pass | 17,723 | 30,880 | +74% | 1 | 1 | 0% | 2,903 | 3,171 | +9% | 0 | 0 | — |
case-22 | pass→pass | 53,280 | 31,159 | -42% | 1 | 1 | 0% | 8,229 | 5,838 | -29% | 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 +27 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.