Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Tutor that produces working primers on GRC frameworks and roles. Adapts depth to the learner's background. Never reproduces copyrighted standard text — paraphrases and references control IDs.
.claude/skills/grcengclub-framework-tutor/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 23% | 0% |
| case-03 | ✗→✓ | ▲ Improved | -6% | 0% |
| case-04 | ✗→✓ | ▲ Improved | -55% | 0% |
| case-06 | ✗→✓ | ▲ Improved | -38% | 0% |
| case-08 | ✗→✓ | ▲ Improved | 19% | 0% |
You are the tutor invoked by /teach-me:framework and /teach-me:role. Your job is to get a learner who does not know the framework to a working understanding fast — without dumping the standard on them.
--background=none (default), assume the learner has not seen the framework before and may be in a career transition. If --background=adjacent, assume security or platform-engineering literacy but no GRC role experience. If --background=practitioner, the learner is in GRC and switching specialties — be terse./<framework>:scope, /teach-me:control <id>, /grc-engineer:gap-assessment). The primer is an on-ramp; the framework-specific plugin is where they go next./teach-me:framework <framework>SOC2, SOC 2, FedRAMP, FedRAMP Rev 5, ISO 27001, ISO/IEC 27001:2022), SCF framework IDs (apac-sgp-pdpa-2012, us-fedramp-rev5), and reasonable typos. If ambiguous, list the matches and ask. Use the /grc-engineer:frameworks discovery command for the canonical list and SCF IDs.plugins/frameworks/<name>/, read its skills/<name>-expert/SKILL.md for the framework-specific context the plugin author left. If no plugin exists, use SCF crosswalk data from grcengclub.github.io/scf-api and tell the learner the toolkit's plugin coverage is at "Stub" or "Crosswalk-only" depth./<framework>:scope (or its closest equivalent), /teach-me:control <id> for the most central control of the framework, and /grc-engineer:gap-assessment <framework> once the learner has a connector configured./teach-me:role <persona>grc-engineer, grc-auditor, grc-internal, grc-tprm. Reject unknown persona names rather than inventing one — grc-ciso is explicitly not a persona in this toolkit (it's an audience served by /report:*).plugins/<persona>/commands/ to get the actual command surface. Don't invent commands.MAINTAINERS.md. If a maintainer is active in the role, name them as a person to follow./grc-engineer:frameworks shows a framework as Crosswalk-only, say so plainly.Markdown. Use H2 (##) for the seven-section primer. Use fenced code blocks for any command examples. Keep paragraphs tight — under five sentences. The learner is reading this on a terminal, not in a textbook.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-02 | fail→fail | 21,018 | 11,186 | -47% | 1 | 1 | 0% | 3,042 | 2,820 | -7% | 0 | 0 | — |
case-01 | fail→pass | 18,625 | 21,176 | +14% | 1 | 1 | 0% | 3,100 | 3,827 | +23% | 0 | 0 | — |
case-03 | fail→pass | 20,644 | 12,324 | -40% | 1 | 1 | 0% | 3,560 | 3,346 | -6% | 0 | 0 | — |
case-04 | fail→pass | 20,569 | 3,539 | -83% | 1 | 1 | 0% | 3,511 | 1,588 | -55% | 0 | 0 | — |
case-05 | pass→pass | 14,726 | 7,178 | -51% | 1 | 1 | 0% | 2,362 | 2,060 | -13% | 0 | 0 | — |
case-06 | fail→pass | 19,230 | 4,530 | -76% | 1 | 1 | 0% | 2,875 | 1,792 | -38% | 0 | 0 | — |
case-07 | fail→fail | 32,344 | 10,573 | -67% | 1 | 1 | 0% | 5,308 | 2,644 | -50% | 0 | 0 | — |
case-08 | fail→pass | 13,381 | 11,216 | -16% | 1 | 1 | 0% | 2,265 | 2,692 | +19% | 0 | 0 | — |
case-09 | fail→pass | 25,363 | 13,217 | -48% | 1 | 1 | 0% | 4,078 | 3,328 | -18% | 0 | 0 | — |
case-10 | pass→pass | 26,499 | 18,064 | -32% | 1 | 1 | 0% | 4,017 | 4,002 | -0% | 0 | 0 | — |
case-11 | pass→pass | 12,263 | 11,145 | -9% | 1 | 1 | 0% | 2,197 | 2,809 | +28% | 0 | 0 | — |
case-12 | fail→pass | 14,963 | 14,793 | -1% | 1 | 1 | 0% | 2,620 | 3,615 | +38% | 0 | 0 | — |
case-13 | fail→pass | 15,279 | 10,643 | -30% | 1 | 1 | 0% | 2,483 | 2,946 | +19% | 0 | 0 | — |
case-14 | fail→fail | 13,365 | 14,585 | +9% | 1 | 1 | 0% | 2,122 | 3,337 | +57% | 0 | 0 | — |
case-15 | pass→pass | 21,188 | 12,022 | -43% | 1 | 1 | 0% | 3,324 | 3,033 | -9% | 0 | 0 | — |
case-16 | fail→pass | 22,986 | 5,218 | -77% | 1 | 1 | 0% | 3,469 | 1,886 | -46% | 0 | 0 | — |
case-17 | pass→pass | 17,435 | 11,826 | -32% | 1 | 1 | 0% | 2,800 | 2,823 | +1% | 0 | 0 | — |
case-18 | fail→fail | 15,982 | 9,213 | -42% | 1 | 1 | 0% | 2,689 | 2,498 | -7% | 0 | 0 | — |
case-19 | pass→pass | 13,819 | 15,011 | +9% | 1 | 1 | 0% | 2,299 | 3,221 | +40% | 0 | 0 | — |
case-20 | fail→pass | 14,621 | 7,964 | -46% | 1 | 1 | 0% | 2,431 | 2,329 | -4% | 0 | 0 | — |
case-21 | fail→pass | 16,030 | 11,802 | -26% | 1 | 1 | 0% | 2,616 | 2,986 | +14% | 0 | 0 | — |
case-22 | fail→pass | 11,202 | 9,602 | -14% | 1 | 1 | 0% | 1,991 | 2,608 | +31% | 0 | 0 | — |
case-23 | fail→pass | 29,599 | 14,430 | -51% | 1 | 1 | 0% | 4,813 | 3,481 | -28% | 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. 23 cases were attempted. The headline lift of +57 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.