Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Load a Spec Kitty agent profile on demand for interactive sessions, including identity, governance scope, boundaries, and initialization.
.claude/skills/priivacy-ai-spk-doctrine-profile-load/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-04 | ✗→✓ | ▲ Improved | -57% | 0% |
| case-06 | ✗→✓ | ▲ Improved | -34% | 0% |
| case-07 | ✗→✓ | ▲ Improved | -60% | 0% |
| case-08 | ✗→✓ | ▲ Improved | -69% | 0% |
| case-05 | ✓→✗ | ▼ Worse | -66% | 0% |
Use this skill when the agent needs a profile outside the runtime loop or the user asks to adopt a specific role.
bash spec-kitty agent profile show <profile-id>
bash spec-kitty charter context --action <action> --json
directive and tactic references, collaboration handoffs, and mode defaults.
spk-run-next for Mission advancement.Do not substitute a raw .agent.yaml read for resolution. A narrowly scoped read-only-harness fallback is documented in the reference below.
ad-hoc-profile-load is a compatibility alias that points here. This skill and its reference are the canonical authority.
references/profile-load-mechanics.md -- Full resolver flow, applicationchecklist, standalone dispatch, and the bounded read-only fallback.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-11 | fail→fail | 9,475 | 4,250 | -55% | 1 | 1 | 0% | 1,716 | 504 | -71% | 0 | 0 | — |
case-01 | fail→fail | 4,277 | 14,136 | +231% | 1 | 1 | 0% | 520 | 565 | +9% | 0 | 0 | — |
case-02 | fail→fail | 6,318 | 14,155 | +124% | 1 | 1 | 0% | 1,043 | 532 | -49% | 0 | 0 | — |
case-03 | fail→fail | 4,569 | 4,378 | -4% | 1 | 1 | 0% | 713 | 538 | -25% | 0 | 0 | — |
case-04 | fail→pass | 10,742 | 2,641 | -75% | 1 | 1 | 0% | 1,863 | 803 | -57% | 0 | 0 | — |
case-05 | pass→fail | 6,988 | 2,648 | -62% | 1 | 1 | 0% | 1,335 | 456 | -66% | 0 | 0 | — |
case-06 | fail→pass | 7,002 | 2,552 | -64% | 1 | 1 | 0% | 1,176 | 774 | -34% | 0 | 0 | — |
case-07 | fail→pass | 7,632 | 1,566 | -79% | 1 | 1 | 0% | 1,340 | 531 | -60% | 0 | 0 | — |
case-08 | fail→pass | 9,343 | 1,277 | -86% | 1 | 1 | 0% | 1,569 | 482 | -69% | 0 | 0 | — |
case-09 | fail→fail | 7,737 | 4,396 | -43% | 1 | 1 | 0% | 1,421 | 570 | -60% | 0 | 0 | — |
case-10 | fail→fail | 4,630 | 4,227 | -9% | 1 | 1 | 0% | 695 | 504 | -27% | 0 | 0 | — |
case-12 | fail→fail | 5,976 | 3,823 | -36% | 1 | 1 | 0% | 1,042 | 507 | -51% | 0 | 0 | — |
case-13 | fail→fail | 12,438 | 2,200 | -82% | 1 | 1 | 0% | 2,218 | 517 | -77% | 0 | 0 | — |
case-14 | fail→fail | 7,333 | 5,509 | -25% | 1 | 1 | 0% | 1,213 | 469 | -61% | 0 | 0 | — |
case-15 | fail→fail | 9,915 | 4,392 | -56% | 1 | 1 | 0% | 1,680 | 558 | -67% | 0 | 0 | — |
case-16 | fail→fail | 8,126 | 14,850 | +83% | 1 | 1 | 0% | 1,392 | 1,810 | +30% | 0 | 0 | — |
case-17 | fail→fail | 10,450 | 12,863 | +23% | 1 | 1 | 0% | 1,653 | 468 | -72% | 0 | 0 | — |
case-18 | pass→pass | 8,471 | 2,706 | -68% | 1 | 1 | 0% | 1,280 | 721 | -44% | 0 | 0 | — |
case-19 | pass→pass | 4,127 | 1,413 | -66% | 1 | 1 | 0% | 736 | 509 | -31% | 0 | 0 | — |
case-20 | fail→fail | 6,374 | 4,332 | -32% | 1 | 1 | 0% | 1,226 | 1,033 | -16% | 0 | 0 | — |
case-21 | pass→fail | 13,150 | 11,045 | -16% | 1 | 1 | 0% | 2,700 | 2,334 | -14% | 0 | 0 | — |
case-22 | pass→pass | 8,815 | 7,645 | -13% | 1 | 1 | 0% | 1,567 | 1,669 | +7% | 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 10 counted toward the lift figure. The other 12 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 +9 percentage points is the difference between those two pass rates over the 10 comparable cases. 2 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.