Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Track a team's budget so surprises die young — the commitment-based view (spent + committed + planned, not just invoiced), the category grain that matches how the team actually spends, the monthly close with its one-sentence read, and the forecast honesty for the year-end question. Use when asked track my team's budget, are we going to blow the budget, why did finance's number surprise us, or set up budget visibility for the team. Produces the three-lane tracker (spent/committed/planned), the mo
.claude/skills/mohitagw15856-team-budget-tracker/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 8% | 0% |
| case-02 | ✗→✓ | ▲ Improved | -1% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 29% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 109% | 0% |
| case-09 | ✗→✓ | ▲ Improved | 69% | 0% |
Team budgets surprise their owners for one structural reason: the official number tracks invoices (what finance has processed), while the truth includes commitments (the signed contract not yet billed, the accepted quote, the hire with an offer out) — so a team can be "under budget" while being catastrophically over-committed. The tracker runs three lanes — spent (invoiced), committed (contractually owed), planned (intended but not yet signed) — against the budget line, at the budget-tracker-design statement-grain, with the monthly close and its one-sentence read. The remaining-to-spend number the team actually needs is budget − spent − committed, and most surprise stories are the word "committed" going untracked.
Ask for these if not provided:
| Category | Budget | Spent | Committed | Planned | True remaining | |---|---|---|---|---|---|
The at-signature rule · the sources (who can commit) · the billing-schedule field]
Reconcile vs. finance → roll billed commitments → the one sentence, written]
Year-end arithmetic with the confidence band · the trajectory signal status · the early-warning conversation if warranted]
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 28,472 | 23,478 | -18% | 1 | 1 | 0% | 4,336 | 4,663 | +8% | 0 | 0 | — |
case-02 | fail→pass | 34,097 | 31,948 | -6% | 1 | 1 | 0% | 5,375 | 5,334 | -1% | 0 | 0 | — |
case-03 | fail→pass | 27,385 | 19,668 | -28% | 1 | 1 | 0% | 3,908 | 5,054 | +29% | 0 | 0 | — |
case-04 | pass→pass | 6,272 | 13,653 | +118% | 1 | 1 | 0% | 1,182 | 2,535 | +114% | 0 | 0 | — |
case-05 | pass→pass | 19,426 | 27,494 | +42% | 1 | 1 | 0% | 3,034 | 5,099 | +68% | 0 | 0 | — |
case-06 | pass→pass | 22,860 | 37,970 | +66% | 1 | 1 | 0% | 4,089 | 9,511 | +133% | 0 | 0 | — |
case-07 | fail→pass | 11,521 | 16,204 | +41% | 1 | 1 | 0% | 1,646 | 3,444 | +109% | 0 | 0 | — |
case-08 | pass→pass | 10,719 | 14,615 | +36% | 1 | 1 | 0% | 1,912 | 3,147 | +65% | 0 | 0 | — |
case-09 | fail→pass | 10,113 | 10,378 | +3% | 1 | 1 | 0% | 1,614 | 2,730 | +69% | 0 | 0 | — |
case-10 | pass→pass | 23,019 | 16,358 | -29% | 1 | 1 | 0% | 3,806 | 4,297 | +13% | 0 | 0 | — |
case-11 | fail→fail | 14,502 | 13,781 | -5% | 1 | 1 | 0% | 1,789 | 3,463 | +94% | 0 | 0 | — |
case-12 | fail→pass | 6,255 | 5,161 | -17% | 1 | 1 | 0% | 1,270 | 2,338 | +84% | 0 | 0 | — |
case-13 | pass→pass | 13,573 | 13,252 | -2% | 1 | 1 | 0% | 1,790 | 3,570 | +99% | 0 | 0 | — |
case-14 | pass→pass | 10,112 | 19,329 | +91% | 1 | 1 | 0% | 1,570 | 4,106 | +162% | 0 | 0 | — |
case-15 | fail→pass | 16,160 | 11,533 | -29% | 1 | 1 | 0% | 2,008 | 3,094 | +54% | 0 | 0 | — |
case-16 | fail→pass | 10,592 | 13,420 | +27% | 1 | 1 | 0% | 1,782 | 3,021 | +70% | 0 | 0 | — |
case-17 | pass→pass | 15,582 | 13,194 | -15% | 1 | 1 | 0% | 2,454 | 3,390 | +38% | 0 | 0 | — |
case-18 | fail→pass | 14,157 | 22,957 | +62% | 1 | 1 | 0% | 2,035 | 3,721 | +83% | 0 | 0 | — |
case-19 | fail→pass | 14,276 | 9,795 | -31% | 1 | 1 | 0% | 2,082 | 2,913 | +40% | 0 | 0 | — |
case-20 | pass→pass | 19,446 | 15,752 | -19% | 1 | 1 | 0% | 1,828 | 3,342 | +83% | 0 | 0 | — |
case-21 | pass→pass | 18,817 | 7,688 | -59% | 1 | 1 | 0% | 1,781 | 2,539 | +43% | 0 | 0 | — |
case-22 | fail→pass | 18,111 | 13,866 | -23% | 1 | 1 | 0% | 2,339 | 3,584 | +53% | 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 +50 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.