Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Legacy alias for resolver-backed profile loading. Use the canonical spk-doctrine-profile-load skill for identity, boundaries, and governance. Triggers: "act as the architect", "load the reviewer profile", "switch to researcher", "use the planner role", "adopt a profile".
.claude/skills/priivacy-ai-ad-hoc-profile-load/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | -58% | 0% |
| case-02 | ✗→✓ | ▲ Improved | 21% | 0% |
| case-07 | ✗→✓ | ▲ Improved | -69% | 0% |
| case-08 | ✗→✓ | ▲ Improved | -47% | 0% |
| case-09 | ✗→✓ | ▲ Improved | -60% | 0% |
This is the compatibility alias for spk-doctrine-profile-load. The canonical skill owns the mechanics; do not maintain a second profile-loading procedure here.
Resolve the requested profile and load action-scoped governance:
bashspec-kitty agent profile show <profile-id> spec-kitty charter context --action <action> --json
Apply the resolved initialization declaration, specialization boundaries, directive and tactic references, collaboration handoffs, and mode defaults.
Read ../spk-doctrine-profile-load/references/profile-load-mechanics.md for the complete resolver-backed flow, including the explicitly degraded fallback for a read-only harness that cannot invoke the CLI.
For a one-shot governed request outside a Mission, use:
bashspec-kitty dispatch "<request>" --profile <profile-id>
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 9,713 | 3,552 | -63% | 1 | 1 | 0% | 1,797 | 752 | -58% | 0 | 0 | — |
case-02 | fail→pass | 7,650 | 6,446 | -16% | 1 | 1 | 0% | 1,102 | 1,329 | +21% | 0 | 0 | — |
case-03 | fail→fail | 11,693 | 7,628 | -35% | 1 | 1 | 0% | 1,477 | 565 | -62% | 0 | 0 | — |
case-04 | pass→pass | 11,626 | 10,284 | -12% | 1 | 1 | 0% | 2,021 | 1,878 | -7% | 0 | 0 | — |
case-05 | pass→fail | 12,294 | 7,533 | -39% | 1 | 1 | 0% | 2,093 | 1,494 | -29% | 0 | 0 | — |
case-06 | pass→pass | 13,871 | 6,595 | -52% | 1 | 1 | 0% | 2,674 | 1,363 | -49% | 0 | 0 | — |
case-07 | fail→pass | 10,236 | 2,261 | -78% | 1 | 1 | 0% | 1,850 | 577 | -69% | 0 | 0 | — |
case-08 | fail→pass | 5,671 | 2,358 | -58% | 1 | 1 | 0% | 992 | 525 | -47% | 0 | 0 | — |
case-09 | fail→pass | 7,027 | 2,163 | -69% | 1 | 1 | 0% | 1,354 | 545 | -60% | 0 | 0 | — |
case-10 | fail→pass | 19,818 | 2,407 | -88% | 1 | 1 | 0% | 1,329 | 603 | -55% | 0 | 0 | — |
case-11 | fail→pass | 6,387 | 2,497 | -61% | 1 | 1 | 0% | 1,083 | 625 | -42% | 0 | 0 | — |
case-12 | fail→pass | 8,960 | 1,933 | -78% | 1 | 1 | 0% | 1,676 | 561 | -67% | 0 | 0 | — |
case-13 | fail→pass | 6,236 | 1,906 | -69% | 1 | 1 | 0% | 1,105 | 536 | -51% | 0 | 0 | — |
case-14 | pass→pass | 5,394 | 1,854 | -66% | 1 | 1 | 0% | 893 | 500 | -44% | 0 | 0 | — |
case-15 | fail→pass | 4,520 | 3,374 | -25% | 1 | 1 | 0% | 785 | 564 | -28% | 0 | 0 | — |
case-16 | pass→pass | 5,502 | 2,829 | -49% | 1 | 1 | 0% | 936 | 624 | -33% | 0 | 0 | — |
case-17 | pass→pass | 5,808 | 1,392 | -76% | 1 | 1 | 0% | 860 | 425 | -51% | 0 | 0 | — |
case-18 | fail→pass | 13,582 | 1,534 | -89% | 1 | 1 | 0% | 2,372 | 460 | -81% | 0 | 0 | — |
case-19 | pass→pass | 4,384 | 2,405 | -45% | 1 | 1 | 0% | 689 | 576 | -16% | 0 | 0 | — |
case-20 | fail→pass | 12,924 | 4,101 | -68% | 1 | 1 | 0% | 1,957 | 927 | -53% | 0 | 0 | — |
case-21 | fail→fail | 12,248 | 2,953 | -76% | 1 | 1 | 0% | 1,890 | 623 | -67% | 0 | 0 | — |
case-22 | fail→pass | 11,772 | 2,534 | -78% | 1 | 1 | 0% | 1,365 | 524 | -62% | 0 | 0 | — |
case-23 | fail→pass | 6,353 | 2,575 | -59% | 1 | 1 | 0% | 996 | 522 | -48% | 0 | 0 | — |
case-24 | pass→pass | 5,256 | 1,659 | -68% | 1 | 1 | 0% | 888 | 426 | -52% | 0 | 0 | — |
case-25 | fail→pass | 13,040 | 1,778 | -86% | 1 | 1 | 0% | 2,089 | 471 | -77% | 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. 25 cases were attempted, and 24 counted toward the lift figure. The other 1 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 +56 percentage points is the difference between those two pass rates over the 24 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.