Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Expertise on OSCAL (Open Security Controls Assessment Language) — what document types exist, when to use each, schema versioning, FedRAMP/eMASS/CSPM integration, round-trip workflows.
.claude/skills/grcengclub-oscal-expert/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 13% | 0% |
| case-14 | ✗→✓ | ▲ Improved | 20% | 0% |
| case-15 | ✗→✓ | ▲ Improved | -5% | 0% |
| case-20 | ✗→✓ | ▲ Improved | -1% | 0% |
| case-21 | ✗→✓ | ▲ Improved | -62% | 0% |
You are the guide for OSCAL document authoring, validation, and conversion in this plugin. Your job:
oscal validate errors and propose fixes./grc-engineer:gap-assessment --output=oscal-ar, the fedramp-ssp plugin, and downstream tools like Compliance Trestle, eMASS, and FedRAMP 20X.OSCAL defines seven document types, all interlinked by UUID references:
catalog → list of controls (e.g., NIST 800-53)
▲
│ imports
profile → baseline (a set of catalog controls + tailoring, e.g., FedRAMP Moderate)
▲
│ imports
ssp → System Security Plan: how *this* system implements the profile
▲
│ imports-ssp
ap → Assessment Plan: what the assessor will test
│
▼
ar → Assessment Results: findings + observations from executing the AP
│
▼
poam → Plan of Action and Milestones: remediation schedule for failed findings
component-definition → reusable "this is how this product implements controls" catalog/grc-engineer:gap-assessment --output=oscal-ar produces AR-shaped output.| Error | Meaning | Fix | |---|---|---| | required property 'uuid' missing | every OSCAL object needs a UUID | generate with uuidgen and insert | | instance type (X) does not match schema type (Y) | wrong data type — usually a string vs array | check schema docs for that field | | enum value not allowed | you've used a value outside the allowed set (e.g. implementation-status) | consult nist.gov/OSCAL/concepts/ for the valid set | | additional property not allowed | FedRAMP/vendor namespacing required | use props with an ns URI, e.g. ns: "https://fedramp.gov/ns/oscal" | | oscal-version mismatch | your document says 1.0 but oscal-cli validates against 1.1.3 | update oscal-version to 1.1.3 |
/grc-engineer:gap-assessment --output=oscal-ar → emits minimal OSCAL Assessment Results. Validate with /oscal:validate; convert to XML for Compliance Trestle with /oscal:convert.fedramp-ssp plugin → takes FedRAMP DOCX SSP templates and produces OSCAL 1.2.0 SSP JSON. Pipe to /oscal:validate before delivering to your FedRAMP PMO.ethanolivertroy/compliance-trestle-skills): IBM's OSCAL authoring toolchain. This plugin's output is designed to round-trip through Trestle.The bundled schema in oscal-cli is 1.1.3. frdocx-to-froscal-ssp produces 1.2.0. These are schema-compatible for the SSP subset used, but declare "oscal-version": "1.1.3" in output if you need strict 1.1.3 compliance, or call /oscal:setup --from-source to get the latest schema bundle.
fedramp-ssp plugin.| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 13,978 | 18,013 | +29% | 1 | 1 | 0% | 2,352 | 2,661 | +13% | 0 | 0 | — |
case-02 | pass→pass | 7,587 | 7,177 | -5% | 1 | 1 | 0% | 1,171 | 2,166 | +85% | 0 | 0 | — |
case-03 | pass→pass | 6,266 | 7,594 | +21% | 1 | 1 | 0% | 1,013 | 2,243 | +121% | 0 | 0 | — |
case-04 | pass→pass | 6,225 | 8,738 | +40% | 1 | 1 | 0% | 975 | 2,463 | +153% | 0 | 0 | — |
case-05 | pass→pass | 3,872 | 4,717 | +22% | 1 | 1 | 0% | 621 | 1,847 | +197% | 0 | 0 | — |
case-06 | pass→pass | 5,312 | 9,465 | +78% | 1 | 1 | 0% | 850 | 2,565 | +202% | 0 | 0 | — |
case-07 | pass→pass | 4,555 | 7,161 | +57% | 1 | 1 | 0% | 754 | 2,213 | +194% | 0 | 0 | — |
case-08 | pass→pass | 5,318 | 5,727 | +8% | 1 | 1 | 0% | 849 | 2,041 | +140% | 0 | 0 | — |
case-09 | pass→pass | 8,092 | 6,313 | -22% | 1 | 1 | 0% | 1,305 | 1,994 | +53% | 0 | 0 | — |
case-10 | pass→pass | 13,199 | 9,464 | -28% | 1 | 1 | 0% | 2,300 | 2,657 | +16% | 0 | 0 | — |
case-11 | pass→pass | 12,334 | 9,316 | -24% | 1 | 1 | 0% | 1,984 | 2,501 | +26% | 0 | 0 | — |
case-12 | pass→pass | 13,167 | 8,021 | -39% | 1 | 1 | 0% | 2,249 | 2,404 | +7% | 0 | 0 | — |
case-13 | pass→pass | 3,035 | 4,451 | +47% | 1 | 1 | 0% | 439 | 1,772 | +304% | 0 | 0 | — |
case-14 | fail→pass | 9,970 | 5,916 | -41% | 1 | 1 | 0% | 1,722 | 2,064 | +20% | 0 | 0 | — |
case-15 | fail→pass | 15,570 | 7,974 | -49% | 1 | 1 | 0% | 2,479 | 2,366 | -5% | 0 | 0 | — |
case-16 | pass→pass | 11,048 | 7,749 | -30% | 1 | 1 | 0% | 1,833 | 2,376 | +30% | 0 | 0 | — |
case-17 | pass→pass | 8,549 | 5,845 | -32% | 1 | 1 | 0% | 1,403 | 1,960 | +40% | 0 | 0 | — |
case-18 | pass→pass | 4,792 | 4,979 | +4% | 1 | 1 | 0% | 738 | 1,738 | +136% | 0 | 0 | — |
case-19 | pass→pass | 4,924 | 4,610 | -6% | 1 | 1 | 0% | 770 | 1,676 | +118% | 0 | 0 | — |
case-20 | fail→pass | 22,617 | 4,887 | -78% | 1 | 1 | 0% | 1,731 | 1,714 | -1% | 0 | 0 | — |
case-21 | fail→pass | 27,734 | 8,307 | -70% | 1 | 1 | 0% | 6,165 | 2,345 | -62% | 0 | 0 | — |
case-22 | fail→pass | 5,093 | 8,305 | +63% | 1 | 1 | 0% | 717 | 2,460 | +243% | 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 +27 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.