opsec-reviewlisted

Red-teams documents, code, READMEs, social posts, commit messages, and paste content for inadvertent operational security leakage before external exposure. Use when user says "review this before I post", "OPSEC check", "is this safe to share", "audit this README", "pre-flight check", "what does this reveal", "check before I push", or when preparing any artifact for public release, GitHub push, social media, client delivery, or external API submission. Distinct from PII redaction — focuses on what an adversary can INFER from context, metadata, structure, and timing signals, not just explicit identifiers.
fevra-dev/Subrosa · ★ 0 · Code & Development · score 72

Install: claude install-skill fevra-dev/Subrosa

# OPSEC Review Adversarial pre-flight audit for artifacts before external exposure. Identifies unintended disclosure through inference, metadata, structure, and content signals — not just explicit sensitive data. **Pair with `redact` skill** for PII/credential stripping. This skill covers what `redact` misses: the inferential layer. --- ## Artifact Types Declare the artifact type at the start of review. Each has a distinct threat surface. | Type | Load reference | |---|---| | Code / repository | `references/code-signals.md` | | README / documentation | `references/doc-signals.md` | | Social post / LinkedIn / blog | `references/social-signals.md` | | Commit message / PR description | `references/doc-signals.md` | | Paste / log dump / stack trace | `references/code-signals.md` | | API payload / schema | `references/code-signals.md` | | General prose / email | `references/doc-signals.md` | If artifact type is ambiguous, identify it from content before proceeding. --- ## Threat Model Reviews are scoped to one of three adversary profiles. Confirm with user or infer from context. ### Profile A — PUBLIC INTERNET Passive adversary. Aggregates public data. Goal: fingerprint stack, infer org structure, identify vulnerabilities, find attack surface. Typical threat: competitor, researcher, automated scanner, bug bounty hunter, recruiter building a dossier. ### Profile B — TARGETED Active adversary with specific interest in the user or their org. Goal: correlate across source