Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Expertise in FedRAMP POA&M lifecycle management, FedRAMP 20x VDR generation, and vulnerability classification using CISA KEV, EPSS, N-ratings, LEV/IRV, and NIST 800-53 control mappings.
.claude/skills/grcengclub-poam-automation-expert/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-07 | ✗→✓ | ▲ Improved | -14% | 0% |
| case-01 | ✗→✓ | ▲ Improved | 163% | 0% |
| case-02 | ✗→✓ | ▲ Improved | 4% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 81% | 0% |
| case-08 | ✗→✓ | ▲ Improved | -17% | 0% |
This connector wraps the FedRAMP POA&M Automation Tool — a Python CLI that manages the full vulnerability finding lifecycle from scanner import through FedRAMP 20x VDR generation.
Each finding is classified using live threat intelligence:
Per FedRAMP 20x baseline:
Findings open 192+ days are automatically moved to the Accepted Vulnerabilities sheet per VDR-TFR-MAV.
This connector maps findings to NIST 800-53 Rev 5 controls via keyword matching across 30+ control families including SI-2, RA-5, AC-2, AC-3, SC-8, SC-28, AU-2, CM-6, CM-7, and others.
pip install openpyxl| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-05 | fail→fail | 14,421 | 10,473 | -27% | 1 | 1 | 0% | 2,440 | 2,510 | +3% | 0 | 0 | — |
case-06 | fail→fail | 17,741 | 14,916 | -16% | 1 | 1 | 0% | 3,630 | 3,846 | +6% | 0 | 0 | — |
case-07 | fail→pass | 17,008 | 9,813 | -42% | 1 | 1 | 0% | 2,684 | 2,299 | -14% | 0 | 0 | — |
case-01 | fail→pass | 8,516 | 18,745 | +120% | 1 | 1 | 0% | 1,264 | 3,318 | +163% | 0 | 0 | — |
case-02 | fail→pass | 17,770 | 15,794 | -11% | 1 | 1 | 0% | 3,258 | 3,394 | +4% | 0 | 0 | — |
case-03 | fail→pass | 15,135 | 19,788 | +31% | 1 | 1 | 0% | 2,659 | 4,803 | +81% | 0 | 0 | — |
case-04 | fail→fail | 16,103 | 15,640 | -3% | 1 | 1 | 0% | 3,017 | 3,751 | +24% | 0 | 0 | — |
case-08 | fail→pass | 17,894 | 9,063 | -49% | 1 | 1 | 0% | 2,807 | 2,335 | -17% | 0 | 0 | — |
case-09 | pass→pass | 7,463 | 2,157 | -71% | 1 | 1 | 0% | 1,231 | 998 | -19% | 0 | 0 | — |
case-10 | pass→pass | 5,866 | 3,611 | -38% | 1 | 1 | 0% | 901 | 1,264 | +40% | 0 | 0 | — |
case-11 | fail→pass | 12,920 | 3,480 | -73% | 1 | 1 | 0% | 2,054 | 1,204 | -41% | 0 | 0 | — |
case-12 | fail→pass | 17,458 | 3,646 | -79% | 1 | 1 | 0% | 2,641 | 1,235 | -53% | 0 | 0 | — |
case-13 | fail→fail | 13,154 | 6,047 | -54% | 1 | 1 | 0% | 1,123 | 1,541 | +37% | 0 | 0 | — |
case-14 | fail→pass | 14,481 | 5,554 | -62% | 1 | 1 | 0% | 2,054 | 1,535 | -25% | 0 | 0 | — |
case-15 | fail→pass | 23,908 | 2,046 | -91% | 1 | 1 | 0% | 1,381 | 944 | -32% | 0 | 0 | — |
case-16 | pass→pass | 12,835 | 5,500 | -57% | 1 | 1 | 0% | 1,997 | 1,485 | -26% | 0 | 0 | — |
case-17 | fail→pass | 7,530 | 14,111 | +87% | 1 | 1 | 0% | 1,041 | 865 | -17% | 0 | 0 | — |
case-18 | fail→pass | 7,134 | 2,323 | -67% | 1 | 1 | 0% | 1,090 | 1,037 | -5% | 0 | 0 | — |
case-19 | fail→pass | 9,212 | 3,434 | -63% | 1 | 1 | 0% | 1,476 | 1,272 | -14% | 0 | 0 | — |
case-20 | fail→pass | 11,212 | 2,011 | -82% | 1 | 1 | 0% | 1,845 | 1,015 | -45% | 0 | 0 | — |
case-21 | pass→pass | 11,908 | 4,164 | -65% | 1 | 1 | 0% | 1,962 | 1,066 | -46% | 0 | 0 | — |
case-22 | fail→pass | 13,681 | 6,132 | -55% | 1 | 1 | 0% | 1,975 | 1,700 | -14% | 0 | 0 | — |
case-23 | fail→pass | 13,315 | 2,173 | -84% | 1 | 1 | 0% | 2,101 | 988 | -53% | 0 | 0 | — |
case-24 | fail→pass | 13,793 | 3,385 | -75% | 1 | 1 | 0% | 2,118 | 1,184 | -44% | 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. 24 cases were attempted, and 23 counted toward the lift figure. The other 1 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 +67 percentage points is the difference between those two pass rates over the 23 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.