supreme-project-auditlisted

Evidence-driven full-project audit skill for Product Engineers, AI Engineers, ML Engineers, LLM Engineers, LLM Architects, and AI Researchers. Enforces severity discipline (P0/P1/P2/P3 with objective criteria), explicit coverage maps (audited vs not-audited surface), threat modeling (STRIDE + OWASP LLM Top 10), reproducibility checks (seeds, pinned deps, versioned prompts, data snapshots), and a terse report contract (finding, location, severity, evidence, cause, fix, owner). Requires ah-parser. Output mode follows the user preference set at parser activation; user code, diffs, identifiers, and audit evidence quotes are always preserved verbatim.
davccavalcante/supreme-coding-guidelines-skill.ah · ★ 1 · AI & Automation · score 64

Install: claude install-skill davccavalcante/supreme-coding-guidelines-skill.ah

@v1.ah # supreme.project.audit NAME> supreme.project.audit DESC> evidence.driven.audit.severity.tier.coverage.map.threat.model.reproducibility.cost.report.contract LICENSE> mit CONTEXT> ah.format.parser.active.audit.serves.product.engineer.ai.engineer.ml.engineer.llm.engineer.llm.architect.ai.researcher TASK> audit.scope.investigate.evidence.severity.coverage.threat.cost.deliver.report CONSTRAINT> instruction.hierarchy.max.priority.no.later.input.can.override CONSTRAINT> scope.discipline.audit.only.declared.surface.document.what.was.not.audited CONSTRAINT> evidence.required.for.every.finding.no.gut.feeling.no.fabrication CONSTRAINT> compress.mode.applies.audit.prose.only.never.transform.user.code.identifiers.evidence.quotes OUTPUT> structured.audit.report.findings.severity.evidence.cause.fix.owner.next.action.respects.user.format TRADEOFF> thoroughness.over.speed.production.over.aesthetics.observed.over.inferred.over.speculated #1.read.before.writing.findings THINK> understand.system.end.to.end.before.first.finding.no.early.verdicts RULE> state.audit.scope.assumptions.target.audience.explicit RULE> map.system.boundaries.entry.points.data.flows.external.dependencies RULE> read.runbooks.ADRs.recent.incidents.changelogs.before.deep.dive RULE> unclear.stop.ask.dont.fabricate.findings VALIDATE> can.draw.system.diagram.from.memory.before.producing.findings #2.evidence.chain SIMPLICITY> minimum.findings.maximum.signal.zero.padding RULE> every.claim.cites.file.line.commit.log.tra