Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Guide for assembling a workspace capability (skills + integrations). Use when the user wants to create a capability.
.claude/skills/holaboss-ai-capability-creator/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-08 | ✗→✓ | ▲ Improved | -41% | 0% |
| case-11 | ✗→✓ | ▲ Improved | -60% | 0% |
| case-12 | ✗→✓ | ▲ Improved | -56% | 0% |
| case-15 | ✗→✓ | ▲ Improved | -55% | 0% |
| case-17 | ✗→✓ | ▲ Improved | -32% | 0% |
A capability bundles one or more skills and the integrations they need into one named, installable unit. Use this when the user wants to create a capability.
skills/<skill-id>/. Create anymissing skill with the skill-creator guidance first.
capabilities/<capability-id>/capability.yaml at theworkspace root. Reference existing skills by ref (do not copy them). Declare each external service the capability needs under integrations.
capability_install tool with capability_id: <capability-id>.needs_connection, tell theuser which one to connect.
yamlid: <capability-id> # must match the directory name name: <Readable Name> description: <one line> version: 0.1.0 skills: - ref: <existing-skill-id> # reference a workspace skill by id integrations: - provider: <provider-id> # e.g. linkedin, gmail, twitter required: true reason: <why this capability needs it> agent_prompt: | <short guidance appended to AGENTS.md describing when/how to use this capability>
Keep the manifest minimal. Prefer ref skills over embedding path skills.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 5,988 | 9,022 | +51% | 1 | 1 | 0% | 838 | 754 | -10% | 0 | 0 | — |
case-02 | fail→fail | 9,488 | 11,739 | +24% | 1 | 1 | 0% | 1,766 | 2,054 | +16% | 0 | 0 | — |
case-03 | fail→fail | 12,714 | 19,372 | +52% | 1 | 1 | 0% | 1,488 | 693 | -53% | 0 | 0 | — |
case-04 | fail→fail | 5,391 | 5,856 | +9% | 1 | 1 | 0% | 320 | 647 | +102% | 0 | 0 | — |
case-05 | pass→pass | 2,936 | 4,847 | +65% | 1 | 1 | 0% | 295 | 1,156 | +292% | 0 | 0 | — |
case-06 | pass→pass | 5,881 | 3,946 | -33% | 1 | 1 | 0% | 1,054 | 992 | -6% | 0 | 0 | — |
case-07 | pass→pass | 5,877 | 5,240 | -11% | 1 | 1 | 0% | 1,127 | 1,244 | +10% | 0 | 0 | — |
case-08 | fail→pass | 8,616 | 2,854 | -67% | 1 | 1 | 0% | 1,485 | 874 | -41% | 0 | 0 | — |
case-09 | pass→pass | 10,048 | 2,898 | -71% | 1 | 1 | 0% | 1,493 | 857 | -43% | 0 | 0 | — |
case-10 | pass→pass | 7,265 | 3,098 | -57% | 1 | 1 | 0% | 1,251 | 993 | -21% | 0 | 0 | — |
case-11 | fail→pass | 13,099 | 3,198 | -76% | 1 | 1 | 0% | 2,395 | 968 | -60% | 0 | 0 | — |
case-12 | fail→pass | 12,265 | 3,441 | -72% | 1 | 1 | 0% | 2,149 | 941 | -56% | 0 | 0 | — |
case-13 | pass→pass | 8,386 | 1,941 | -77% | 1 | 1 | 0% | 901 | 647 | -28% | 0 | 0 | — |
case-14 | pass→pass | 7,618 | 2,692 | -65% | 1 | 1 | 0% | 1,091 | 827 | -24% | 0 | 0 | — |
case-15 | fail→pass | 7,977 | 1,731 | -78% | 1 | 1 | 0% | 1,285 | 579 | -55% | 0 | 0 | — |
case-16 | pass→pass | 6,260 | 1,255 | -80% | 1 | 1 | 0% | 1,021 | 566 | -45% | 0 | 0 | — |
case-17 | fail→pass | 6,146 | 2,441 | -60% | 1 | 1 | 0% | 1,054 | 720 | -32% | 0 | 0 | — |
case-18 | fail→pass | 7,353 | 5,089 | -31% | 1 | 1 | 0% | 1,337 | 1,378 | +3% | 0 | 0 | — |
case-19 | pass→fail | 10,579 | 2,160 | -80% | 1 | 1 | 0% | 1,783 | 701 | -61% | 0 | 0 | — |
case-20 | pass→pass | 14,129 | 2,942 | -79% | 1 | 1 | 0% | 2,709 | 929 | -66% | 0 | 0 | — |
case-21 | pass→pass | 5,589 | 2,523 | -55% | 1 | 1 | 0% | 1,097 | 800 | -27% | 0 | 0 | — |
case-22 | fail→pass | 10,814 | 4,342 | -60% | 1 | 1 | 0% | 1,674 | 1,175 | -30% | 0 | 0 | — |
case-23 | fail→pass | 6,106 | 6,935 | +14% | 1 | 1 | 0% | 1,147 | 1,751 | +53% | 0 | 0 | — |
case-24 | pass→pass | 29,290 | 1,800 | -94% | 1 | 1 | 0% | 4,678 | 598 | -87% | 0 | 0 | — |
case-25 | fail→pass | 6,257 | 4,418 | -29% | 1 | 1 | 0% | 1,101 | 1,212 | +10% | 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. 25 cases were attempted, and 21 counted toward the lift figure. The other 4 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 21 comparable cases. 2 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.