Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Decode a veterinary treatment estimate — what each line is for, which items are core vs precautionary, and how to have the options conversation nobody offers you. Use when someone asks 'is this vet estimate reasonable', 'decode my vet's treatment plan', 'do we need all these tests', or 'I can't afford this vet bill what are my options'. Produces a line-by-line decode with core/precautionary/comfort triage, the questions that surface the tiered options vets keep in reserve, and the payment and as
.claude/skills/mohitagw15856-vet-estimate-decoder/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 69% | 0% |
| case-02 | ✗→✓ | ▲ Improved | 57% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 64% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 31% | 0% |
| case-08 | ✗→✓ | ▲ Improved | 65% | 0% |
Vet estimates arrive at the worst possible moment — with a sick animal in the room and love doing the negotiating. Most clinics genuinely have tiered options ("gold standard" down to "reasonable and safe"), but the estimate shows only the top tier unless someone asks. This skill decodes what each line does, which items drive the diagnosis versus round it out, and scripts the options conversation — respectfully, because the vet is an ally, not an adversary. It decodes the estimate; it never practices medicine.
Ask for these only if they aren't already provided:
The load-bearing question for every line is: "What decision does this test/treatment inform, and what happens if we stage it?" Staging (treat the likely thing, escalate if no improvement) is a legitimate medical strategy vets use constantly — but usually only when asked. The script's magic sentence: "We want to do right by her and we have a real budget — if this were your animal and money mattered, what would the plan look like?" Vets answer that question honestly and gratefully almost every time.
1. The verdict — total (low–high), the core subset, and the two questions most worth asking, in three sentences.
2. Line-by-line decode
| Line | What it's for | Triage (to confirm with vet) | Price | |---|---|---|---|
3. Questions for the vet — 4–6, each opening a real option: "What would the staged version look like?", "Which of these change today's treatment?", "Is outpatient with a recheck viable?", "What's the must-do subset if we're prioritizing?"
4. The options script — the budget-honest conversation, word for word, including the magic sentence.
5. Money paths — clinic payment plans, assistance programs (breed/condition/region-specific ones exist — list types to search, don't invent names), care-credit-style financing decoded (deferred-interest traps flagged), insurance claim steps if covered.
End the artifact with, verbatim: "This is a plain-language reading, not legal/financial advice — laws vary by jurisdiction; confirm anything load-bearing with a qualified professional."
Client-side veterinary cost-conversation practice — line triage, staged-diagnostics questioning, tiered-plan elicitation.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 17,610 | 35,611 | +102% | 1 | 1 | 0% | 3,331 | 5,620 | +69% | 0 | 0 | — |
case-02 | fail→pass | 29,180 | 31,003 | +6% | 1 | 1 | 0% | 3,841 | 6,016 | +57% | 0 | 0 | — |
case-03 | fail→pass | 36,579 | 20,873 | -43% | 1 | 1 | 0% | 2,858 | 4,673 | +64% | 0 | 0 | — |
case-04 | pass→pass | 19,618 | 14,234 | -27% | 1 | 1 | 0% | 2,139 | 3,158 | +48% | 0 | 0 | — |
case-05 | fail→fail | 38,923 | 26,354 | -32% | 1 | 1 | 0% | 3,254 | 4,355 | +34% | 0 | 0 | — |
case-06 | fail→fail | 24,105 | 29,634 | +23% | 1 | 1 | 0% | 2,904 | 4,862 | +67% | 0 | 0 | — |
case-07 | fail→pass | 26,194 | 25,235 | -4% | 1 | 1 | 0% | 2,896 | 3,803 | +31% | 0 | 0 | — |
case-08 | fail→pass | 23,859 | 29,765 | +25% | 1 | 1 | 0% | 3,029 | 4,990 | +65% | 0 | 0 | — |
case-09 | pass→pass | 20,257 | 28,950 | +43% | 1 | 1 | 0% | 2,043 | 4,537 | +122% | 0 | 0 | — |
case-10 | fail→pass | 24,623 | 33,316 | +35% | 1 | 1 | 0% | 2,572 | 4,338 | +69% | 0 | 0 | — |
case-11 | pass→pass | 18,188 | 36,315 | +100% | 1 | 1 | 0% | 2,067 | 3,936 | +90% | 0 | 0 | — |
case-12 | fail→pass | 20,516 | 30,596 | +49% | 1 | 1 | 0% | 2,483 | 4,335 | +75% | 0 | 0 | — |
case-13 | pass→pass | 22,392 | 31,757 | +42% | 1 | 1 | 0% | 2,519 | 5,079 | +102% | 0 | 0 | — |
case-14 | pass→pass | 17,115 | 21,810 | +27% | 1 | 1 | 0% | 1,485 | 4,562 | +207% | 0 | 0 | — |
case-15 | fail→pass | 17,740 | 24,659 | +39% | 1 | 1 | 0% | 2,144 | 4,294 | +100% | 0 | 0 | — |
case-16 | pass→pass | 16,008 | 22,418 | +40% | 1 | 1 | 0% | 2,023 | 4,153 | +105% | 0 | 0 | — |
case-17 | pass→pass | 17,611 | 16,572 | -6% | 1 | 1 | 0% | 1,994 | 3,881 | +95% | 0 | 0 | — |
case-18 | pass→pass | 13,695 | 21,135 | +54% | 1 | 1 | 0% | 2,114 | 4,239 | +101% | 0 | 0 | — |
case-19 | pass→pass | 22,118 | 33,070 | +50% | 1 | 1 | 0% | 2,623 | 5,117 | +95% | 0 | 0 | — |
case-20 | pass→pass | 20,303 | 23,768 | +17% | 1 | 1 | 0% | 2,054 | 4,372 | +113% | 0 | 0 | — |
case-21 | pass→pass | 16,377 | 19,739 | +21% | 1 | 1 | 0% | 1,780 | 3,538 | +99% | 0 | 0 | — |
case-22 | pass→pass | 23,942 | 25,107 | +5% | 1 | 1 | 0% | 2,208 | 4,396 | +99% | 0 | 0 | — |
case-23 | fail→pass | 13,669 | 25,584 | +87% | 1 | 1 | 0% | 1,466 | 4,039 | +176% | 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. 23 cases were attempted. The headline lift of +39 percentage points is the difference between those two pass rates over the 23 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.