Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Write a team's working agreements — the small set of explicit norms (communication, meetings, decisions, conflict) that replace the assumptions people were silently violating, built from the team's actual frictions and revisited on a cadence. Use when asked create team working agreements, set norms for our new team, we keep clashing over how we work, or onboard people into how this team operates. Produces the friction-derived agreement set, the specific-behavior phrasing, the disagreement protoc
.claude/skills/mohitagw15856-working-agreements/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 21% | 0% |
| case-02 | ✗→✓ | ▲ Improved | 27% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 26% | 0% |
| case-04 | ✗→✓ | ▲ Improved | 22% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 73% | 0% |
Teams run on invisible rules — response-time expectations, camera norms, how decisions happen, whether 6pm messages demand answers — and every clash between two people's invisible rules gets misread as a character flaw ("he's unresponsive," "she's pushy"). Working agreements make the rules visible: a small set (6–10, not a constitution), derived from the team's actual frictions (not copied from a blog), phrased as observable behaviors ("we respond to non-urgent messages within one working day" — checkable) rather than values ("we communicate respectfully" — vibes), and revisited quarterly, because teams change and dead agreements poison live ones.
Ask for these if not provided:
| # | Agreement (observable) | The friction it resolves | |---|---|---| 6–10 · the disagreement protocol among them, always]
Openly, where · the escalation path · the commit norm]
Where it's pinned · the newcomer walkthrough · the quarterly two-question revisit, calendared]
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 21,715 | 30,332 | +40% | 1 | 1 | 0% | 3,450 | 4,174 | +21% | 0 | 0 | — |
case-02 | fail→pass | 28,500 | 27,067 | -5% | 1 | 1 | 0% | 3,444 | 4,373 | +27% | 0 | 0 | — |
case-03 | fail→pass | 28,990 | 29,451 | +2% | 1 | 1 | 0% | 3,365 | 4,249 | +26% | 0 | 0 | — |
case-04 | fail→pass | 29,981 | 26,502 | -12% | 1 | 1 | 0% | 3,749 | 4,580 | +22% | 0 | 0 | — |
case-05 | pass→pass | 24,244 | 25,119 | +4% | 1 | 1 | 0% | 2,825 | 3,971 | +41% | 0 | 0 | — |
case-06 | pass→fail | 21,958 | 34,814 | +59% | 1 | 1 | 0% | 2,611 | 5,118 | +96% | 0 | 0 | — |
case-07 | fail→pass | 19,284 | 22,079 | +14% | 1 | 1 | 0% | 2,136 | 3,688 | +73% | 0 | 0 | — |
case-08 | fail→pass | 33,930 | 22,831 | -33% | 1 | 1 | 0% | 4,569 | 3,777 | -17% | 0 | 0 | — |
case-09 | fail→fail | 17,800 | 24,318 | +37% | 1 | 1 | 0% | 2,384 | 3,701 | +55% | 0 | 0 | — |
case-10 | fail→pass | 16,799 | 26,592 | +58% | 1 | 1 | 0% | 1,657 | 4,300 | +160% | 0 | 0 | — |
case-11 | fail→pass | 18,401 | 16,388 | -11% | 1 | 1 | 0% | 2,003 | 3,587 | +79% | 0 | 0 | — |
case-12 | fail→pass | 22,942 | 24,475 | +7% | 1 | 1 | 0% | 2,665 | 4,313 | +62% | 0 | 0 | — |
case-13 | fail→pass | 10,887 | 23,802 | +119% | 1 | 1 | 0% | 1,786 | 3,916 | +119% | 0 | 0 | — |
case-14 | fail→pass | 23,848 | 26,039 | +9% | 1 | 1 | 0% | 2,852 | 4,402 | +54% | 0 | 0 | — |
case-15 | fail→pass | 21,791 | 23,836 | +9% | 1 | 1 | 0% | 2,870 | 3,937 | +37% | 0 | 0 | — |
case-16 | pass→pass | 21,520 | 26,151 | +22% | 1 | 1 | 0% | 2,352 | 4,212 | +79% | 0 | 0 | — |
case-17 | fail→pass | 22,292 | 18,038 | -19% | 1 | 1 | 0% | 2,496 | 3,067 | +23% | 0 | 0 | — |
case-18 | pass→pass | 24,300 | 20,559 | -15% | 1 | 1 | 0% | 3,130 | 4,124 | +32% | 0 | 0 | — |
case-19 | fail→pass | 23,209 | 21,283 | -8% | 1 | 1 | 0% | 2,739 | 3,566 | +30% | 0 | 0 | — |
case-20 | fail→pass | 19,482 | 22,476 | +15% | 1 | 1 | 0% | 2,100 | 3,846 | +83% | 0 | 0 | — |
case-21 | fail→pass | 14,866 | 22,546 | +52% | 1 | 1 | 0% | 2,175 | 3,664 | +68% | 0 | 0 | — |
case-22 | fail→pass | 18,213 | 14,324 | -21% | 1 | 1 | 0% | 2,798 | 3,068 | +10% | 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 +73 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.