Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Use this guidance when an AI agent is about to add or update dependencies.
.claude/skills/hashgraph-online-hol-guard-skill-guidance/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | -68% | 0% |
| case-03 | ✗→✓ | ▲ Improved | -42% | 0% |
| case-05 | ✗→✓ | ▲ Improved | -59% | 0% |
| case-06 | ✗→✓ | ▲ Improved | -61% | 0% |
| case-07 | ✗→✓ | ▲ Improved | -61% | 0% |
Use this guidance when an AI agent is about to add or update dependencies.
hol-guard protect --dry-run -- npm install <package>hol-guard supply-chain scan --jsonhol-guard supply-chain explain <package>@<version> --ecosystem <ecosystem>hol-guard package-shims status --jsonhol-guard package-shims repair --manager npm --jsonCursor has two protection surfaces. Cursor editor covers MCP servers in .cursor/mcp.json. Cursor CLI covers the cursor-agent command path. Keep them distinct when activating, checking, repairing, or removing Guard.
hol-guard apps connect cursor --surface editorhol-guard apps connect cursor --surface clihol-guard apps test cursor --surface editorhol-guard apps test cursor --surface clihol-guard apps repair cursor --surface editorhol-guard apps repair cursor --surface clihol-guard apps disconnect cursor --surface editor --confirm disconnect-cursorhol-guard apps disconnect cursor --surface cli --confirm disconnect-cursorGuard owns trust checks, drift repair, redacted receipts, and Cloud sync. Cursor owns its native editor and CLI behavior. If a surface is missing, unsupported, or unavailable, report that state instead of inventing an install URL or fallback command.
hol-guard protect -- npm cihol-guard supply-chain audit --json| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 39,622 | 2,675 | -93% | 1 | 1 | 0% | 3,147 | 1,020 | -68% | 0 | 0 | — |
case-02 | fail→fail | 18,798 | 3,513 | -81% | 1 | 1 | 0% | 1,821 | 998 | -45% | 0 | 0 | — |
case-03 | fail→pass | 15,752 | 5,261 | -67% | 1 | 1 | 0% | 2,525 | 1,461 | -42% | 0 | 0 | — |
case-04 | pass→pass | 14,718 | 2,961 | -80% | 1 | 1 | 0% | 2,743 | 786 | -71% | 0 | 0 | — |
case-05 | fail→pass | 12,053 | 2,421 | -80% | 1 | 1 | 0% | 2,000 | 822 | -59% | 0 | 0 | — |
case-06 | fail→pass | 13,732 | 3,961 | -71% | 1 | 1 | 0% | 2,260 | 883 | -61% | 0 | 0 | — |
case-07 | fail→pass | 16,975 | 3,277 | -81% | 1 | 1 | 0% | 2,663 | 1,034 | -61% | 0 | 0 | — |
case-08 | fail→pass | 14,119 | 3,953 | -72% | 1 | 1 | 0% | 2,077 | 1,220 | -41% | 0 | 0 | — |
case-09 | fail→pass | 11,579 | 2,250 | -81% | 1 | 1 | 0% | 1,559 | 848 | -46% | 0 | 0 | — |
case-10 | fail→pass | 9,168 | 2,098 | -77% | 1 | 1 | 0% | 1,541 | 818 | -47% | 0 | 0 | — |
case-11 | pass→pass | 10,807 | 3,188 | -71% | 1 | 1 | 0% | 1,554 | 1,039 | -33% | 0 | 0 | — |
case-12 | pass→pass | 6,124 | 2,978 | -51% | 1 | 1 | 0% | 903 | 828 | -8% | 0 | 0 | — |
case-13 | pass→pass | 13,292 | 4,602 | -65% | 1 | 1 | 0% | 2,077 | 1,252 | -40% | 0 | 0 | — |
case-14 | pass→pass | 15,012 | 3,462 | -77% | 1 | 1 | 0% | 2,467 | 1,098 | -55% | 0 | 0 | — |
case-15 | fail→pass | 5,566 | 1,714 | -69% | 1 | 1 | 0% | 877 | 815 | -7% | 0 | 0 | — |
case-16 | fail→pass | 4,882 | 3,284 | -33% | 1 | 1 | 0% | 654 | 687 | +5% | 0 | 0 | — |
case-17 | fail→pass | 19,965 | 3,537 | -82% | 1 | 1 | 0% | 3,818 | 1,091 | -71% | 0 | 0 | — |
case-18 | pass→pass | 9,131 | 2,621 | -71% | 1 | 1 | 0% | 1,476 | 956 | -35% | 0 | 0 | — |
case-19 | pass→pass | 8,690 | 2,419 | -72% | 1 | 1 | 0% | 1,390 | 812 | -42% | 0 | 0 | — |
case-20 | pass→pass | 8,581 | 6,320 | -26% | 1 | 1 | 0% | 1,470 | 1,431 | -3% | 0 | 0 | — |
case-21 | pass→pass | 14,702 | 6,090 | -59% | 1 | 1 | 0% | 2,500 | 1,524 | -39% | 0 | 0 | — |
case-22 | pass→pass | 12,511 | 9,983 | -20% | 1 | 1 | 0% | 2,167 | 2,215 | +2% | 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 +50 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.