Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Set up the team decision log that ends relitigation — the one-line-per-decision format (what, why, who, when, reopening rule), the capture moments wired into existing rituals, and the lookup habit that makes it pay. Use when asked set up a decision log, we keep re-deciding the same things, where do decisions get recorded, or new people keep asking why we do X. Produces the log format, the capture wiring, the reopening rule, and the retrieval habits.
.claude/skills/mohitagw15856-decision-log-setup/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 5% | 0% |
| case-02 | ✗→✓ | ▲ Improved | 41% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 15% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 63% | 0% |
| case-12 | ✗→✓ | ▲ Improved | 63% | 0% |
Teams re-decide because decisions evaporate: made in a meeting, mentioned in a thread, remembered differently by everyone, and relitigated the moment someone new (or someone persistent) asks "wait, why do we…?" The decision log is the cheapest institutional memory there is — one line per decision (what, why, who, when, and the reopening rule) — but logs fail as shelf-ware unless capture is wired into existing moments (the meeting's last five minutes, the thread's landing) and retrieval becomes reflex ("checked the log?" as the first response to why-questions).
Ask for these if not provided:
| Date | Decision | Why (incl. rejected alt.) | Decider | Reopens if | |---|---|---|---|---|
Venue → moment → who writes the line · the weekly sweep as the catch-all]
The default · how specific triggers get set · the link-instead-of-debate response]
The why-question → link reflex · onboarding's greatest-hits tour · the back-fill list (10–15, marked reconstructed)]
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 21,085 | 14,010 | -34% | 1 | 1 | 0% | 3,279 | 3,447 | +5% | 0 | 0 | — |
case-02 | fail→pass | 18,085 | 20,178 | +12% | 1 | 1 | 0% | 3,163 | 4,464 | +41% | 0 | 0 | — |
case-03 | fail→pass | 21,443 | 17,323 | -19% | 1 | 1 | 0% | 3,547 | 4,084 | +15% | 0 | 0 | — |
case-04 | pass→pass | 23,756 | 18,270 | -23% | 1 | 1 | 0% | 4,197 | 4,270 | +2% | 0 | 0 | — |
case-05 | pass→pass | 10,848 | 17,762 | +64% | 1 | 1 | 0% | 2,062 | 4,256 | +106% | 0 | 0 | — |
case-06 | pass→fail | 12,628 | 15,762 | +25% | 1 | 1 | 0% | 2,558 | 3,857 | +51% | 0 | 0 | — |
case-11 | pass→pass | 12,558 | 13,859 | +10% | 1 | 1 | 0% | 2,106 | 3,505 | +66% | 0 | 0 | — |
case-07 | fail→pass | 13,825 | 15,037 | +9% | 1 | 1 | 0% | 2,272 | 3,696 | +63% | 0 | 0 | — |
case-08 | fail→fail | 9,850 | 11,985 | +22% | 1 | 1 | 0% | 1,759 | 3,069 | +74% | 0 | 0 | — |
case-09 | pass→pass | 14,485 | 12,784 | -12% | 1 | 1 | 0% | 2,433 | 3,173 | +30% | 0 | 0 | — |
case-10 | pass→pass | 12,614 | 9,575 | -24% | 1 | 1 | 0% | 2,094 | 2,801 | +34% | 0 | 0 | — |
case-12 | fail→pass | 15,614 | 19,057 | +22% | 1 | 1 | 0% | 2,547 | 4,163 | +63% | 0 | 0 | — |
case-13 | fail→pass | 14,994 | 13,131 | -12% | 1 | 1 | 0% | 2,523 | 3,369 | +34% | 0 | 0 | — |
case-14 | pass→pass | 12,262 | 9,082 | -26% | 1 | 1 | 0% | 1,917 | 2,750 | +43% | 0 | 0 | — |
case-15 | fail→pass | 10,186 | 7,316 | -28% | 1 | 1 | 0% | 1,716 | 2,438 | +42% | 0 | 0 | — |
case-16 | fail→pass | 9,673 | 6,483 | -33% | 1 | 1 | 0% | 1,630 | 2,308 | +42% | 0 | 0 | — |
case-17 | pass→pass | 13,468 | 11,411 | -15% | 1 | 1 | 0% | 2,339 | 3,307 | +41% | 0 | 0 | — |
case-18 | fail→pass | 11,171 | 11,511 | +3% | 1 | 1 | 0% | 2,061 | 3,161 | +53% | 0 | 0 | — |
case-19 | pass→pass | 12,116 | 9,409 | -22% | 1 | 1 | 0% | 2,125 | 2,653 | +25% | 0 | 0 | — |
case-20 | pass→pass | 11,549 | 10,265 | -11% | 1 | 1 | 0% | 2,073 | 2,937 | +42% | 0 | 0 | — |
case-21 | pass→pass | 13,950 | 14,527 | +4% | 1 | 1 | 0% | 2,355 | 3,595 | +53% | 0 | 0 | — |
case-22 | fail→pass | 11,628 | 8,101 | -30% | 1 | 1 | 0% | 1,995 | 2,624 | +32% | 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 +41 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.