Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Simulate the client who grinds on your price — the budget theater, the competitor quote, the scope squeeze — against your actual offer, with a debrief on where you caved and what holding would have sounded like. Use when asked simulate a client negotiating my rate, practice price pushback, they said I'm too expensive, or stress-test my pricing conversation. Produces the negotiation transcript with the client's private playbook notes, the deal outcome, and a debrief on every concession with its s
.claude/skills/mohitagw15856-the-price-pushback/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 8% | 0% |
| case-02 | ✗→✓ | ▲ Improved | 34% | 0% |
| case-06 | ✗→✓ | ▲ Improved | 72% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 71% | 0% |
| case-09 | ✗→✓ | ▲ Improved | 179% | 0% |
Every freelancer loses the same negotiation for years before noticing it's the same negotiation: the sharp intake of breath, the mysterious budget ceiling, the competitor who's "half your price," the request to "just trim the number, same scope." This skill plays the client running that playbook against your real offer — each move labeled in their private notes — then debriefs every point where you softened, with the exact words that would have held. The goal isn't winning every deal; it's never again negotiating against yourself.
Ask for these if not provided:
| Move | What it sounds like | What it actually is | |---|---|---| | The Flinch | "Wow. That's… a lot." + silence | Theater; tests whether you'll negotiate against yourself unprompted | | The Phantom Budget | "We only have $X allocated" | A number chosen after hearing yours; rarely load-bearing | | The Competitor Quote | "Someone quoted half that" | Sometimes real, never comparable — scope is the reply, not price | | The Scope Squeeze | "Same deliverables, sharper price?" | The ask to donate margin; the counter is price-per-scope, always | | The Future Carrot | "Lots more work after this one" | Paying for volume that hasn't been ordered; discounts follow commitments, not promises | | The Deadline Squeeze | "Need your best number by 5pm" | Urgency manufactured to prevent exactly the thinking you're doing now |
Negotiation mechanics to honor: the first person to re-cut their own number loses the frame · silence after stating price is a move, not a malfunction — the simulation holds it uncomfortably long · every concession without an exchange resets the client's target · "let me re-scope to meet that budget" preserves rate while moving price — it's the master response and the simulation rewards it · walking away from below-floor work is a win condition, and one run should end there.
> Simulation — a plausible adversarial reading, not a prediction.
The negotiation, grounded in the supplied offer. Client's note: after each move, naming the play and reading your response — "flinch worked, they offered 10% unprompted."]
Terms reached or walk-away · the client's notes on what they'd actually have paid · what precedent this outcome sets for the next negotiation with them]
| Moment you softened | The move that caused it | The hold (exact words) | |---|---|---| Plus: the re-scope response drafted for this offer · the floor restated with its arithmetic · the silence drill — what 10 seconds of it costs vs. buys]
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 44,854 | 41,281 | -8% | 1 | 1 | 0% | 5,204 | 5,618 | +8% | 0 | 0 | — |
case-02 | fail→pass | 40,689 | 37,386 | -8% | 1 | 1 | 0% | 4,618 | 6,202 | +34% | 0 | 0 | — |
case-03 | pass→fail | 25,792 | 32,276 | +25% | 1 | 1 | 0% | 2,805 | 5,375 | +92% | 0 | 0 | — |
case-04 | pass→fail | 31,324 | 39,382 | +26% | 1 | 1 | 0% | 2,844 | 5,616 | +97% | 0 | 0 | — |
case-05 | pass→pass | 20,324 | 26,206 | +29% | 1 | 1 | 0% | 2,361 | 4,141 | +75% | 0 | 0 | — |
case-06 | fail→pass | 22,142 | 27,798 | +26% | 1 | 1 | 0% | 2,731 | 4,706 | +72% | 0 | 0 | — |
case-07 | fail→pass | 23,477 | 35,619 | +52% | 1 | 1 | 0% | 2,737 | 4,687 | +71% | 0 | 0 | — |
case-08 | fail→fail | 17,985 | 16,981 | -6% | 1 | 1 | 0% | 1,667 | 3,090 | +85% | 0 | 0 | — |
case-09 | fail→pass | 23,119 | 45,408 | +96% | 1 | 1 | 0% | 2,311 | 6,449 | +179% | 0 | 0 | — |
case-10 | fail→pass | 25,337 | 32,991 | +30% | 1 | 1 | 0% | 2,754 | 5,033 | +83% | 0 | 0 | — |
case-11 | fail→pass | 22,550 | 39,583 | +76% | 1 | 1 | 0% | 2,656 | 5,839 | +120% | 0 | 0 | — |
case-12 | fail→pass | 22,358 | 27,016 | +21% | 1 | 1 | 0% | 2,479 | 4,397 | +77% | 0 | 0 | — |
case-13 | fail→fail | 26,244 | 35,553 | +35% | 1 | 1 | 0% | 2,952 | 5,545 | +88% | 0 | 0 | — |
case-14 | fail→fail | 30,919 | 30,922 | +0% | 1 | 1 | 0% | 3,285 | 5,162 | +57% | 0 | 0 | — |
case-15 | fail→fail | 26,703 | 35,195 | +32% | 1 | 1 | 0% | 4,453 | 5,652 | +27% | 0 | 0 | — |
case-16 | fail→pass | 21,983 | 33,323 | +52% | 1 | 1 | 0% | 2,377 | 4,829 | +103% | 0 | 0 | — |
case-17 | fail→pass | 13,660 | 39,471 | +189% | 1 | 1 | 0% | 1,269 | 5,578 | +340% | 0 | 0 | — |
case-18 | fail→pass | 9,454 | 36,403 | +285% | 1 | 1 | 0% | 1,626 | 5,195 | +219% | 0 | 0 | — |
case-19 | pass→pass | 23,149 | 29,232 | +26% | 1 | 1 | 0% | 2,535 | 5,117 | +102% | 0 | 0 | — |
case-20 | fail→pass | 15,088 | 39,796 | +164% | 1 | 1 | 0% | 1,334 | 6,135 | +360% | 0 | 0 | — |
case-21 | fail→pass | 14,939 | 39,102 | +162% | 1 | 1 | 0% | 1,309 | 6,134 | +369% | 0 | 0 | — |
case-22 | fail→pass | 18,446 | 17,709 | -4% | 1 | 1 | 0% | 2,032 | 2,980 | +47% | 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 +55 percentage points is the difference between those two pass rates over the 22 comparable cases. 3 cases got worse with the skill loaded, and they are 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.