Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →FedRAMP Rev 5 authorization expert. Provides guidance on traditional authorization paths, SSP/SAP/SAR/POA&M documentation, NIST 800-53 Rev 5 control implementation, and 3PAO assessment preparation.
.claude/skills/grcengclub-fedramp-rev5-expert/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-15 | ✗→✓ | ▲ Improved | 43% | 0% |
| case-16 | ✗→✓ | ▲ Improved | 47% | 0% |
| case-19 | ✓→✗ | ▼ Worse | -1% | 0% |
| case-07 | ✓→✓ | ＝ Same ✓ | 4% | 0% |
| case-01 | ✓→✓ | ＝ Same ✓ | -8% | 0% |
Deep expertise in traditional FedRAMP authorization under Rev 5.
| Document | Purpose | |----------|---------| | SSP | System Security Plan - control implementation | | SAP | Security Assessment Plan - test procedures | | SAR | Security Assessment Report - findings | | POA&M | Plan of Action & Milestones - remediation |
All 20 control families apply based on baseline:
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-07 | pass→pass | 13,759 | 12,094 | -12% | 1 | 1 | 0% | 2,026 | 2,100 | +4% | 0 | 0 | — |
case-01 | pass→pass | 5,764 | 4,548 | -21% | 1 | 1 | 0% | 901 | 826 | -8% | 0 | 0 | — |
case-02 | pass→pass | 4,778 | 5,244 | +10% | 1 | 1 | 0% | 724 | 1,006 | +39% | 0 | 0 | — |
case-20 | pass→pass | 11,525 | 9,737 | -16% | 1 | 1 | 0% | 1,866 | 1,923 | +3% | 0 | 0 | — |
case-03 | pass→pass | 7,262 | 11,925 | +64% | 1 | 1 | 0% | 1,142 | 2,116 | +85% | 0 | 0 | — |
case-04 | fail→fail | 8,931 | 8,951 | +0% | 1 | 1 | 0% | 1,447 | 1,699 | +17% | 0 | 0 | — |
case-05 | pass→pass | 10,666 | 7,632 | -28% | 1 | 1 | 0% | 1,644 | 1,510 | -8% | 0 | 0 | — |
case-06 | pass→pass | 6,332 | 3,823 | -40% | 1 | 1 | 0% | 1,083 | 917 | -15% | 0 | 0 | — |
case-08 | pass→pass | 8,621 | 6,794 | -21% | 1 | 1 | 0% | 1,449 | 1,343 | -7% | 0 | 0 | — |
case-09 | pass→pass | 9,292 | 9,934 | +7% | 1 | 1 | 0% | 1,376 | 1,757 | +28% | 0 | 0 | — |
case-10 | pass→pass | 8,221 | 7,848 | -5% | 1 | 1 | 0% | 1,189 | 1,529 | +29% | 0 | 0 | — |
case-11 | pass→pass | 9,567 | 7,777 | -19% | 1 | 1 | 0% | 1,462 | 1,510 | +3% | 0 | 0 | — |
case-12 | pass→pass | 11,345 | 11,268 | -1% | 1 | 1 | 0% | 1,671 | 1,998 | +20% | 0 | 0 | — |
case-13 | pass→pass | 12,817 | 13,588 | +6% | 1 | 1 | 0% | 1,821 | 2,283 | +25% | 0 | 0 | — |
case-14 | pass→pass | 7,200 | 4,771 | -34% | 1 | 1 | 0% | 1,177 | 1,086 | -8% | 0 | 0 | — |
case-15 | fail→pass | 5,478 | 5,227 | -5% | 1 | 1 | 0% | 793 | 1,135 | +43% | 0 | 0 | — |
case-16 | fail→pass | 7,994 | 8,860 | +11% | 1 | 1 | 0% | 1,142 | 1,679 | +47% | 0 | 0 | — |
case-17 | pass→pass | 6,222 | 5,644 | -9% | 1 | 1 | 0% | 1,110 | 1,342 | +21% | 0 | 0 | — |
case-18 | pass→pass | 9,332 | 7,507 | -20% | 1 | 1 | 0% | 1,566 | 1,523 | -3% | 0 | 0 | — |
case-19 | pass→fail | 12,820 | 10,925 | -15% | 1 | 1 | 0% | 2,154 | 2,132 | -1% | 0 | 0 | — |
case-21 | pass→pass | 13,278 | 11,846 | -11% | 1 | 1 | 0% | 2,043 | 2,137 | +5% | 0 | 0 | — |
case-22 | pass→pass | 5,276 | 4,964 | -6% | 1 | 1 | 0% | 903 | 1,120 | +24% | 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 +5 percentage points is the difference between those two pass rates over the 22 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.