Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Digest of what your Superset agents did — sweeps workspaces, tasks, and agent terminals, then reports what finished, what needs review, and what's blocked. Use when the user asks what their agents did, wants a standup or summary of agent work, or returns after being away.
.claude/skills/superset-sh-standup/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-05 | ✗→✓ | ▲ Improved | 272% | 0% |
| case-07 | ✗→✓ | ▲ Improved | -60% | 0% |
| case-10 | ✗→✓ | ▲ Improved | -35% | 0% |
| case-11 | ✗→✓ | ▲ Improved | -58% | 0% |
| case-12 | ✗→✓ | ▲ Improved | -13% | 0% |
Answer "what happened while I was away?" from real state, not guesses. Entirely read-only.
superset workspaces list — active workspacessuperset tasks list — task statessuperset terminals list --workspace <id>, then superset terminals read --workspace <id> --terminal <terminalId> for each agent terminal — the last screen of output shows whether the agent finished, asked a question, or erroredLead with what needs the user, one line per item: workspace, agent, state, and the next action. Then in-flight, then completed, then stale-workspace cleanup suggestions. Keep the whole digest scannable — no terminal dumps; quote at most the single relevant line an agent printed.
Never send input to a terminal, modify tasks, or clean anything up as part of the digest — offer those as follow-ups and act only when asked.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-04 | pass→fail | 8,822 | 5,765 | -35% | 1 | 1 | 0% | 1,429 | 535 | -63% | 0 | 0 | — |
case-01 | fail→fail | 16,016 | 6,663 | -58% | 1 | 1 | 0% | 2,536 | 553 | -78% | 0 | 0 | — |
case-02 | fail→fail | 6,536 | 5,824 | -11% | 1 | 1 | 0% | 459 | 532 | +16% | 0 | 0 | — |
case-03 | fail→fail | 14,601 | 6,869 | -53% | 1 | 1 | 0% | 852 | 739 | -13% | 0 | 0 | — |
case-05 | fail→pass | 8,907 | 7,377 | -17% | 1 | 1 | 0% | 438 | 1,628 | +272% | 0 | 0 | — |
case-06 | fail→fail | 7,978 | 12,313 | +54% | 1 | 1 | 0% | 1,380 | 2,018 | +46% | 0 | 0 | — |
case-07 | fail→pass | 14,907 | 3,853 | -74% | 1 | 1 | 0% | 2,348 | 949 | -60% | 0 | 0 | — |
case-08 | pass→pass | 10,134 | 1,358 | -87% | 1 | 1 | 0% | 1,668 | 558 | -67% | 0 | 0 | — |
case-09 | pass→pass | 8,174 | 1,854 | -77% | 1 | 1 | 0% | 1,386 | 627 | -55% | 0 | 0 | — |
case-10 | fail→pass | 7,265 | 2,781 | -62% | 1 | 1 | 0% | 1,150 | 750 | -35% | 0 | 0 | — |
case-11 | fail→pass | 9,448 | 2,001 | -79% | 1 | 1 | 0% | 1,441 | 609 | -58% | 0 | 0 | — |
case-12 | fail→pass | 4,528 | 3,965 | -12% | 1 | 1 | 0% | 701 | 610 | -13% | 0 | 0 | — |
case-13 | pass→pass | 8,550 | 2,036 | -76% | 1 | 1 | 0% | 1,322 | 583 | -56% | 0 | 0 | — |
case-14 | pass→pass | 8,252 | 2,107 | -74% | 1 | 1 | 0% | 1,330 | 631 | -53% | 0 | 0 | — |
case-15 | pass→pass | 8,263 | 2,892 | -65% | 1 | 1 | 0% | 1,311 | 784 | -40% | 0 | 0 | — |
case-16 | fail→pass | 10,013 | 2,448 | -76% | 1 | 1 | 0% | 1,614 | 706 | -56% | 0 | 0 | — |
case-17 | fail→pass | 11,358 | 3,047 | -73% | 1 | 1 | 0% | 1,663 | 809 | -51% | 0 | 0 | — |
case-18 | pass→pass | 11,185 | 2,791 | -75% | 1 | 1 | 0% | 1,647 | 667 | -60% | 0 | 0 | — |
case-19 | fail→fail | 12,899 | 3,987 | -69% | 1 | 1 | 0% | 2,121 | 1,065 | -50% | 0 | 0 | — |
case-20 | fail→pass | 5,218 | 2,236 | -57% | 1 | 1 | 0% | 761 | 604 | -21% | 0 | 0 | — |
case-21 | pass→pass | 10,530 | 2,475 | -76% | 1 | 1 | 0% | 1,570 | 670 | -57% | 0 | 0 | — |
case-22 | pass→pass | 10,148 | 2,323 | -77% | 1 | 1 | 0% | 1,577 | 604 | -62% | 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, and 17 counted toward the lift figure. The other 5 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. The headline lift of +32 percentage points is the difference between those two pass rates over the 17 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.