Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Create or update Aside skills. Use when the user wants to turn reusable instructions, site instructions, workflows, domain knowledge, scripts, references, or templates into an Aside skill under the current agent's skills directory.
.claude/skills/kunanonj-aside-skill-creator/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 36% | 0% |
| case-06 | ✗→✓ | ▲ Improved | 22% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 121% | 0% |
| case-13 | ✗→✓ | ▲ Improved | -48% | 0% |
| case-14 | ✗→✓ | ▲ Improved | 2% | 0% |
Use this skill to create or update Aside skills for the current agent.
Create user-owned skills under:
text<agent-dir>/skills/<skill-name>/SKILL.md
Use bash tool and write_file tool to create the skill file.
Aside reads SKILL.md frontmatter for discovery and activation. Use only fields that Aside understands:
yaml--- name: my-skill description: Clear trigger description for when Aside should use this skill. ---
Write the body as durable operating instructions for a future Aside agent.
Keep the body concise and procedural:
scripts/, references/, or assets/ only when they materially help.Put trigger information in description, not in a "When to use" body section. The body is only useful after the skill has already been selected or loaded.
<agent-dir>/skills/user/<skill-name>/SKILL.md.SKILL.md.| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 12,968 | 14,287 | +10% | 1 | 1 | 0% | 1,709 | 2,323 | +36% | 0 | 0 | — |
case-02 | fail→fail | 12,926 | 4,993 | -61% | 1 | 1 | 0% | 2,931 | 625 | -79% | 0 | 0 | — |
case-03 | fail→fail | 11,619 | 3,319 | -71% | 1 | 1 | 0% | 2,566 | 538 | -79% | 0 | 0 | — |
case-04 | pass→pass | 2,265 | 2,524 | +11% | 1 | 1 | 0% | 465 | 837 | +80% | 0 | 0 | — |
case-05 | pass→pass | 6,005 | 3,680 | -39% | 1 | 1 | 0% | 1,402 | 1,344 | -4% | 0 | 0 | — |
case-06 | fail→pass | 6,521 | 6,132 | -6% | 1 | 1 | 0% | 1,480 | 1,809 | +22% | 0 | 0 | — |
case-07 | fail→pass | 7,825 | 15,191 | +94% | 1 | 1 | 0% | 1,635 | 3,609 | +121% | 0 | 0 | — |
case-08 | fail→fail | 6,917 | 6,936 | +0% | 1 | 1 | 0% | 1,541 | 944 | -39% | 0 | 0 | — |
case-09 | fail→fail | 7,746 | 3,546 | -54% | 1 | 1 | 0% | 1,922 | 621 | -68% | 0 | 0 | — |
case-10 | fail→fail | 12,408 | 4,972 | -60% | 1 | 1 | 0% | 2,861 | 708 | -75% | 0 | 0 | — |
case-11 | fail→fail | 14,365 | 4,815 | -66% | 1 | 1 | 0% | 3,446 | 834 | -76% | 0 | 0 | — |
case-12 | fail→fail | 5,380 | 5,200 | -3% | 1 | 1 | 0% | 1,161 | 840 | -28% | 0 | 0 | — |
case-13 | fail→pass | 7,659 | 2,336 | -69% | 1 | 1 | 0% | 1,745 | 915 | -48% | 0 | 0 | — |
case-14 | fail→pass | 10,319 | 9,195 | -11% | 1 | 1 | 0% | 2,303 | 2,351 | +2% | 0 | 0 | — |
case-15 | pass→pass | 9,164 | 10,796 | +18% | 1 | 1 | 0% | 2,044 | 2,184 | +7% | 0 | 0 | — |
case-16 | pass→pass | 9,103 | 6,619 | -27% | 1 | 1 | 0% | 1,980 | 1,769 | -11% | 0 | 0 | — |
case-17 | pass→pass | 10,511 | 5,589 | -47% | 1 | 1 | 0% | 2,070 | 1,559 | -25% | 0 | 0 | — |
case-18 | fail→fail | 4,704 | 3,451 | -27% | 1 | 1 | 0% | 987 | 578 | -41% | 0 | 0 | — |
case-23 | pass→fail | 7,382 | 5,020 | -32% | 1 | 1 | 0% | 1,486 | 807 | -46% | 0 | 0 | — |
case-19 | fail→fail | 5,828 | 4,656 | -20% | 1 | 1 | 0% | 1,272 | 1,439 | +13% | 0 | 0 | — |
case-20 | fail→fail | 9,323 | 14,066 | +51% | 1 | 1 | 0% | 1,831 | 2,328 | +27% | 0 | 0 | — |
case-21 | fail→fail | 11,137 | 6,259 | -44% | 1 | 1 | 0% | 2,690 | 781 | -71% | 0 | 0 | — |
case-22 | pass→pass | 6,192 | 2,199 | -64% | 1 | 1 | 0% | 1,200 | 857 | -29% | 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, and 13 counted toward the lift figure. The other 10 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 +17 percentage points is the difference between those two pass rates over the 13 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.