Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Personalize COG Second Brain workflow through role pack selection and vault initialization
.claude/skills/a5c-ai-cog-onboarding/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-04 | ✗→✓ | ▲ Improved | -29% | 0% |
| case-10 | ✗→✓ | ▲ Improved | -23% | 0% |
| case-11 | ✗→✓ | ▲ Improved | -14% | 0% |
| case-17 | ✗→✓ | ▲ Improved | 11% | 0% |
| case-18 | ✗→✓ | ▲ Improved | 4% | 0% |
file-read to check for existing vault at the specified pathdirectory-create to build the COG directory structurefile-write to create profile.md in 00-inbox with role pack configurationgit-init to initialize Git repository in the vaultfile-write to create .gitignore with privacy-sensitive patternsgit-commit to commit initial vault structurejson{ "userName": "Alex", "rolePack": "engineer", "vaultPath": "./cog-vault", "integrations": { "github": { "org": "my-org", "repos": ["main-repo"] } } }
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 15,289 | 7,216 | -53% | 1 | 1 | 0% | 3,275 | 925 | -72% | 0 | 0 | — |
case-02 | fail→fail | 15,271 | 6,366 | -58% | 1 | 1 | 0% | 3,167 | 717 | -77% | 0 | 0 | — |
case-03 | fail→fail | 25,715 | 7,615 | -70% | 1 | 1 | 0% | 5,153 | 965 | -81% | 0 | 0 | — |
case-04 | fail→pass | 5,821 | 2,744 | -53% | 1 | 1 | 0% | 1,034 | 738 | -29% | 0 | 0 | — |
case-05 | fail→fail | 10,885 | 6,576 | -40% | 1 | 1 | 0% | 2,152 | 740 | -66% | 0 | 0 | — |
case-06 | fail→fail | 11,730 | 8,491 | -28% | 1 | 1 | 0% | 1,862 | 631 | -66% | 0 | 0 | — |
case-07 | fail→fail | 6,417 | 6,433 | +0% | 1 | 1 | 0% | 1,034 | 853 | -18% | 0 | 0 | — |
case-08 | fail→fail | 5,316 | 2,659 | -50% | 1 | 1 | 0% | 709 | 764 | +8% | 0 | 0 | — |
case-09 | fail→fail | 10,566 | 4,149 | -61% | 1 | 1 | 0% | 1,799 | 973 | -46% | 0 | 0 | — |
case-10 | fail→pass | 6,980 | 3,873 | -45% | 1 | 1 | 0% | 1,274 | 987 | -23% | 0 | 0 | — |
case-11 | fail→pass | 9,302 | 8,065 | -13% | 1 | 1 | 0% | 1,730 | 1,484 | -14% | 0 | 0 | — |
case-12 | fail→fail | 8,251 | 1,944 | -76% | 1 | 1 | 0% | 1,204 | 584 | -51% | 0 | 0 | — |
case-13 | pass→pass | 30,366 | 1,727 | -94% | 1 | 1 | 0% | 1,812 | 524 | -71% | 0 | 0 | — |
case-14 | fail→fail | 16,932 | 7,377 | -56% | 1 | 1 | 0% | 2,922 | 787 | -73% | 0 | 0 | — |
case-15 | fail→fail | 5,956 | 6,389 | +7% | 1 | 1 | 0% | 318 | 659 | +107% | 0 | 0 | — |
case-16 | fail→fail | 6,080 | 1,910 | -69% | 1 | 1 | 0% | 963 | 551 | -43% | 0 | 0 | — |
case-17 | fail→pass | 3,312 | 2,205 | -33% | 1 | 1 | 0% | 585 | 647 | +11% | 0 | 0 | — |
case-18 | fail→pass | 3,918 | 2,041 | -48% | 1 | 1 | 0% | 608 | 632 | +4% | 0 | 0 | — |
case-19 | fail→pass | 7,600 | 1,930 | -75% | 1 | 1 | 0% | 1,225 | 552 | -55% | 0 | 0 | — |
case-20 | fail→pass | 8,110 | 2,110 | -74% | 1 | 1 | 0% | 1,345 | 611 | -55% | 0 | 0 | — |
case-21 | fail→fail | 11,068 | 10,384 | -6% | 1 | 1 | 0% | 2,019 | 1,883 | -7% | 0 | 0 | — |
case-22 | fail→fail | 15,451 | 7,613 | -51% | 1 | 1 | 0% | 2,868 | 1,828 | -36% | 0 | 0 | — |
case-23 | fail→fail | 11,190 | 8,822 | -21% | 1 | 1 | 0% | 2,544 | 2,069 | -19% | 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 15 counted toward the lift figure. The other 8 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 +30 percentage points is the difference between those two pass rates over the 15 comparable cases. 5 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.