Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Switch model profile for kata agents (quality/balanced/budget). Triggers include "set profile", "set profile".
.claude/skills/majiayu000-kata-set-profile/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-05 | ✗→✓ | ▲ Improved | 7% | 0% |
| case-09 | ✗→✓ | ▲ Improved | 356% | 0% |
| case-10 | ✗→✓ | ▲ Improved | -23% | 0% |
| case-11 | ✗→✓ | ▲ Improved | -49% | 0% |
| case-12 | ✗→✓ | ▲ Improved | -44% | 0% |
<objective> Switch the model profile used by Kata agents. This controls which Claude model each agent uses, balancing quality vs token spend. </objective>
<profiles> | Profile | Description | | ------------ | -------------------------------------------------------------- | | quality | Opus everywhere except read-only verification | | balanced | Opus for planning, Sonnet for execution/verification (default) | | budget | Sonnet for writing, Haiku for research/verification | </profiles>
<process>
if $ARGUMENTS.profile not in ["quality", "balanced", "budget"]:
Error: Invalid profile "$ARGUMENTS.profile"
Valid profiles: quality, balanced, budget
STOPbashls .planning/config.json 2>/dev/null
If no .planning/ directory:
Error: No Kata project found.
Run /kata-new-project first to initialize a project.Read current config:
bashcat .planning/config.json
Update model_profile field (or add if missing):
json{ "model_profile": "$ARGUMENTS.profile" }
Write updated config back to .planning/config.json.
✓ Model profile set to: $ARGUMENTS.profile
Agents will now use:
[Show table from model-profiles.md for selected profile]
Next spawned agents will use the new profile.</process>
<examples>
Switch to budget mode:
/kata-set-profile budget
✓ Model profile set to: budget
Agents will now use:
| Agent | Model |
| ------------- | ------ |
| kata-planner | sonnet |
| kata-executor | sonnet |
| kata-verifier | haiku |
| ... | ... |Switch to quality mode:
/kata-set-profile quality
✓ Model profile set to: quality
Agents will now use:
| Agent | Model |
| ------------- | ------ |
| kata-planner | opus |
| kata-executor | opus |
| kata-verifier | sonnet |
| ... | ... |</examples>
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 14,704 | 15,816 | +8% | 1 | 1 | 0% | 210 | 838 | +299% | 0 | 0 | — |
case-07 | fail→fail | 9,753 | 5,131 | -47% | 1 | 1 | 0% | 360 | 794 | +121% | 0 | 0 | — |
case-02 | fail→fail | 8,941 | 9,525 | +7% | 1 | 1 | 0% | 1,404 | 753 | -46% | 0 | 0 | — |
case-03 | fail→fail | 14,374 | 6,389 | -56% | 1 | 1 | 0% | 1,550 | 729 | -53% | 0 | 0 | — |
case-04 | fail→fail | 6,750 | 12,100 | +79% | 1 | 1 | 0% | 1,134 | 956 | -16% | 0 | 0 | — |
case-05 | fail→pass | 7,969 | 10,273 | +29% | 1 | 1 | 0% | 1,055 | 1,127 | +7% | 0 | 0 | — |
case-06 | fail→fail | 12,055 | 5,679 | -53% | 1 | 1 | 0% | 678 | 846 | +25% | 0 | 0 | — |
case-08 | fail→fail | 4,930 | 11,155 | +126% | 1 | 1 | 0% | 676 | 913 | +35% | 0 | 0 | — |
case-09 | fail→pass | 6,026 | 6,455 | +7% | 1 | 1 | 0% | 242 | 1,104 | +356% | 0 | 0 | — |
case-10 | fail→pass | 9,705 | 4,720 | -51% | 1 | 1 | 0% | 1,551 | 1,192 | -23% | 0 | 0 | — |
case-11 | fail→pass | 10,885 | 2,258 | -79% | 1 | 1 | 0% | 1,870 | 945 | -49% | 0 | 0 | — |
case-12 | fail→pass | 13,430 | 3,614 | -73% | 1 | 1 | 0% | 2,073 | 1,170 | -44% | 0 | 0 | — |
case-13 | fail→fail | 2,813 | 6,332 | +125% | 1 | 1 | 0% | 301 | 831 | +176% | 0 | 0 | — |
case-14 | fail→pass | 16,819 | 2,119 | -87% | 1 | 1 | 0% | 1,752 | 811 | -54% | 0 | 0 | — |
case-15 | fail→fail | 2,626 | 2,003 | -24% | 1 | 1 | 0% | 205 | 820 | +300% | 0 | 0 | — |
case-16 | pass→fail | 8,924 | 5,361 | -40% | 1 | 1 | 0% | 1,444 | 746 | -48% | 0 | 0 | — |
case-17 | fail→pass | 15,449 | 2,705 | -82% | 1 | 1 | 0% | 2,027 | 971 | -52% | 0 | 0 | — |
case-18 | fail→pass | 12,218 | 3,625 | -70% | 1 | 1 | 0% | 1,835 | 994 | -46% | 0 | 0 | — |
case-19 | pass→pass | 13,415 | 3,059 | -77% | 1 | 1 | 0% | 2,107 | 1,062 | -50% | 0 | 0 | — |
case-20 | fail→pass | 5,979 | 9,077 | +52% | 1 | 1 | 0% | 194 | 2,053 | +958% | 0 | 0 | — |
case-21 | pass→pass | 12,200 | 15,714 | +29% | 1 | 1 | 0% | 2,456 | 3,441 | +40% | 0 | 0 | — |
case-22 | pass→fail | 9,103 | 7,458 | -18% | 1 | 1 | 0% | 1,151 | 1,077 | -6% | 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 +32 percentage points is the difference between those two pass rates over the 10 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.