Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Manage credentials in OS keychains across Windows, macOS, and Linux
.claude/skills/a5c-ai-keychain-credential-manager/SKILL.md| Model | Eval pass | Runs |
|---|---|---|
| gemini-3.6-flash | 100% | 71 |
| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | -13% | 0% |
| case-02 | ✗→✓ | ▲ Improved | 11% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 158% | 0% |
| case-05 | ✗→✓ | ▲ Improved | -65% | 0% |
| case-06 | ✗→✓ | ▲ Improved | -61% | 0% |
Manage credentials securely in OS keychains (Windows Credential Manager, macOS Keychain, Linux libsecret) with a cross-platform API.
json{ "type": "object", "properties": { "projectPath": { "type": "string" }, "serviceName": { "type": "string" }, "framework": { "enum": ["electron", "native", "dotnet"] } }, "required": ["projectPath", "serviceName"] }
javascript// Using keytar (Electron) const keytar = require('keytar'); async function storeCredential(account, password) { await keytar.setPassword('MyApp', account, password); } async function getCredential(account) { return await keytar.getPassword('MyApp', account); }
security-hardening processelectron-ipc-security-audit| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 18,891 | 11,324 | -40% | 1 | 1 | 0% | 3,100 | 2,686 | -13% | 0 | 0 | — |
case-02 | fail→pass | 9,504 | 10,976 | +15% | 1 | 1 | 0% | 1,792 | 1,982 | +11% | 0 | 0 | — |
case-03 | fail→pass | 6,737 | 17,315 | +157% | 1 | 1 | 0% | 1,490 | 3,849 | +158% | 0 | 0 | — |
case-04 | pass→pass | 9,223 | 9,920 | +8% | 1 | 1 | 0% | 1,701 | 585 | -66% | 0 | 0 | — |
case-05 | fail→pass | 8,914 | 2,828 | -68% | 1 | 1 | 0% | 1,693 | 588 | -65% | 0 | 0 | — |
case-06 | fail→pass | 7,462 | 1,736 | -77% | 1 | 1 | 0% | 1,480 | 583 | -61% | 0 | 0 | — |
case-07 | pass→pass | 5,227 | 4,529 | -13% | 1 | 1 | 0% | 853 | 1,200 | +41% | 0 | 0 | — |
case-08 | pass→pass | 8,020 | 3,637 | -55% | 1 | 1 | 0% | 1,187 | 1,030 | -13% | 0 | 0 | — |
case-09 | pass→pass | 4,644 | 3,234 | -30% | 1 | 1 | 0% | 827 | 792 | -4% | 0 | 0 | — |
case-10 | pass→pass | 12,751 | 5,551 | -56% | 1 | 1 | 0% | 2,131 | 1,143 | -46% | 0 | 0 | — |
case-11 | pass→pass | 7,059 | 3,251 | -54% | 1 | 1 | 0% | 1,071 | 702 | -34% | 0 | 0 | — |
case-12 | fail→pass | 7,529 | 7,954 | +6% | 1 | 1 | 0% | 1,341 | 1,899 | +42% | 0 | 0 | — |
case-13 | fail→pass | 14,025 | 2,404 | -83% | 1 | 1 | 0% | 2,197 | 670 | -70% | 0 | 0 | — |
case-14 | fail→pass | 13,345 | 1,691 | -87% | 1 | 1 | 0% | 2,380 | 471 | -80% | 0 | 0 | — |
case-15 | fail→pass | 13,230 | 11,148 | -16% | 1 | 1 | 0% | 2,662 | 2,306 | -13% | 0 | 0 | — |
case-16 | pass→pass | 5,705 | 3,397 | -40% | 1 | 1 | 0% | 980 | 753 | -23% | 0 | 0 | — |
case-17 | fail→pass | 9,606 | 4,532 | -53% | 1 | 1 | 0% | 1,558 | 535 | -66% | 0 | 0 | — |
case-18 | pass→pass | 10,765 | 7,435 | -31% | 1 | 1 | 0% | 1,683 | 1,416 | -16% | 0 | 0 | — |
case-19 | pass→pass | 7,250 | 5,327 | -27% | 1 | 1 | 0% | 1,207 | 1,317 | +9% | 0 | 0 | — |
case-20 | pass→pass | 7,669 | 5,405 | -30% | 1 | 1 | 0% | 1,494 | 1,277 | -15% | 0 | 0 | — |
case-21 | pass→pass | 5,216 | 2,684 | -49% | 1 | 1 | 0% | 935 | 661 | -29% | 0 | 0 | — |
case-22 | pass→pass | 13,823 | 13,125 | -5% | 1 | 1 | 0% | 2,685 | 2,797 | +4% | 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. The headline lift of +45 percentage points is the difference between those two pass rates over the 22 comparable cases.
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.