Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Read this skill when the user uses 1Password.
.claude/skills/kunanonj-aside-1password/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-10 | ✗→✓ | ▲ Improved | -64% | 0% |
| case-04 | ✗→✓ | ▲ Improved | -33% | 0% |
| case-05 | ✗→✓ | ▲ Improved | -29% | 0% |
| case-06 | ✗→✓ | ▲ Improved | -14% | 0% |
| case-07 | ✗→✓ | ▲ Improved | -52% | 0% |
1Password autofill options appear as an iframe popover menu after focusing a form field.
Then check whether the bottom of the snapshot includes iframe [origin="1Password"].
status: "1Password menu is available. Press down arrow to select." is shown, 1Password is locked. Follow the unlock flow below.Open the 1Password extension page only to unlock 1Password with Aside PWM:
chrome-extension://aeblfdkhhhdcdjpifhhbdiojplfjncoa/popup/index.html
Then run:
jsawait passwordManager.unlockExternalPasswordManager(page, '1password');
After unlocking 1Password:
Do not use the 1Password extension page as the primary place to search and autofill website logins. Use the original sign-in page's 1Password autofill menu for website login autofill.
If Aside PWM is locked, ask the user to unlock either Aside PWM or 1Password.
If no saved 1Password unlock item is available in Aside PWM, fall back to other password manager options. Ask the user to unlock 1Password directly only as the last resort.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-10 | fail→pass | 12,351 | 3,967 | -68% | 1 | 1 | 0% | 2,268 | 810 | -64% | 0 | 0 | — |
case-01 | fail→fail | 4,197 | 8,080 | +93% | 1 | 1 | 0% | 649 | 1,041 | +60% | 0 | 0 | — |
case-02 | fail→fail | 5,692 | 9,784 | +72% | 1 | 1 | 0% | 912 | 633 | -31% | 0 | 0 | — |
case-03 | fail→fail | 6,432 | 9,557 | +49% | 1 | 1 | 0% | 414 | 584 | +41% | 0 | 0 | — |
case-04 | fail→pass | 9,357 | 3,271 | -65% | 1 | 1 | 0% | 1,499 | 1,008 | -33% | 0 | 0 | — |
case-05 | fail→pass | 7,273 | 4,272 | -41% | 1 | 1 | 0% | 996 | 706 | -29% | 0 | 0 | — |
case-06 | fail→pass | 13,986 | 8,915 | -36% | 1 | 1 | 0% | 980 | 839 | -14% | 0 | 0 | — |
case-07 | fail→pass | 9,115 | 1,758 | -81% | 1 | 1 | 0% | 1,655 | 787 | -52% | 0 | 0 | — |
case-08 | pass→pass | 4,305 | 4,890 | +14% | 1 | 1 | 0% | 727 | 762 | +5% | 0 | 0 | — |
case-09 | pass→pass | 10,123 | 2,596 | -74% | 1 | 1 | 0% | 1,569 | 799 | -49% | 0 | 0 | — |
case-11 | fail→pass | 4,230 | 3,466 | -18% | 1 | 1 | 0% | 716 | 842 | +18% | 0 | 0 | — |
case-12 | fail→pass | 8,187 | 2,562 | -69% | 1 | 1 | 0% | 1,413 | 947 | -33% | 0 | 0 | — |
case-13 | pass→pass | 9,566 | 4,889 | -49% | 1 | 1 | 0% | 1,863 | 714 | -62% | 0 | 0 | — |
case-14 | fail→pass | 9,347 | 2,864 | -69% | 1 | 1 | 0% | 1,852 | 771 | -58% | 0 | 0 | — |
case-19 | fail→pass | 11,642 | 3,662 | -69% | 1 | 1 | 0% | 1,703 | 948 | -44% | 0 | 0 | — |
case-15 | pass→pass | 9,260 | 1,517 | -84% | 1 | 1 | 0% | 1,554 | 693 | -55% | 0 | 0 | — |
case-16 | pass→pass | 10,034 | 4,315 | -57% | 1 | 1 | 0% | 1,554 | 776 | -50% | 0 | 0 | — |
case-17 | pass→pass | 5,911 | 3,747 | -37% | 1 | 1 | 0% | 916 | 736 | -20% | 0 | 0 | — |
case-18 | pass→pass | 14,644 | 5,229 | -64% | 1 | 1 | 0% | 2,158 | 975 | -55% | 0 | 0 | — |
case-20 | pass→pass | 21,250 | 13,434 | -37% | 1 | 1 | 0% | 1,708 | 1,462 | -14% | 0 | 0 | — |
case-21 | pass→pass | 9,616 | 4,390 | -54% | 1 | 1 | 0% | 1,725 | 1,126 | -35% | 0 | 0 | — |
case-22 | pass→pass | 14,460 | 3,860 | -73% | 1 | 1 | 0% | 2,520 | 1,153 | -54% | 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 19 counted toward the lift figure. The other 3 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 +41 percentage points is the difference between those two pass rates over the 19 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.