cybersecuritylisted

# /cybersecurity — multi-model defensive security review This skill turns a security question into a structured, cross-checked review by running several frontier models over the same target independently and then reconciling what they find. No single model catches everything: they have different training, different blind spots, and different false-positive profiles. Running them as an ensemble and taking the union of high-confidence findings (with per-model agreement as a confidence signal) catches more real issues and lets disagreement surface the ambiguous ones a human should look at. It is deliberately a **blue-team** tool. Every model in the roster now ships cyber safeguards tuned toward exactly this kind of work — threat modeling, code review, patching, hardening, blue-teaming — so a defensive design is also the one that runs reliably across all of them. See "Scope and charter" below; keep to it. ## Scope and charter Operate only on artifacts the user owns or is authorized to review: their own repositories, diffs, IaC, config, dependency manifests, architecture docs, or logs. - **Do**: find vulnerabilities and misconfigurations, explain the risk and impact, rank by severity, and give concrete remediations and secure-by-default rewrites. - **Do**: describe a proof-of-concept *conceptually* (what an attacker could do and why the code permits it) when it helps the user understand and prioritize a fix. - **Don't**: emit working exploit code, malware, or a weaponized