界面已本地化;技能介绍与说明文档按所选语言由 AI 自动翻译,代码、命令、路径和专业术语保留原文,以原作者文档为准。
探索技能
开发与工程 · SKILL.md

cyber-framework-mapping

cyber-framework-mapping

【开发与工程】Maintains sparse, reviewable cybersecurity framework edges with versioned IDs, relation, rationale, source, and review status. Use when mapping exercise observations or procedures to NIST, ATT&CK, or another named framework.

独立 GitHub 快照 · 自动分类 · 以原作者文档为准

技能说明

GitHub

name: cyber-framework-mapping guardrail: true description: Maintains sparse, reviewable cybersecurity framework edges with versioned IDs, relation, rationale, source, and review status. Use when mapping exercise observations or procedures to NIST, ATT&CK, or another named framework. metadata: labels: [cybersecurity, framework-mapping, provenance] triggers: keywords: [framework mapping, control mapping, ATT&CK mapping, NIST mapping, framework edge, mapping review]

Cyber Framework Mapping

Priority: P1 (HIGH)

Map only evidenced relationships; catalog context is not proof of efficacy, compliance, or coverage.

Structure

cyber-framework-mapping/
├── SKILL.md
├── references/edge-record.md
└── evals/evals.json

Workflow

  1. Name framework, version, stable ID, and exact source URL or document section.
  2. Choose relation (supports, describes, observed-as, partial, unknown, or conflicts).
  3. Write a bounded rationale tied to an evidence record, not a generic similarity.
  4. Set review status: reviewed, needs-review, or unreviewed; include reviewer and date when reviewed.
  5. Preserve empty or unknown mappings explicitly; do not infer a complete catalog.
  6. On a version/source change, preserve the prior versioned edge and its evidence unchanged. Create a separately versioned candidate linked to the prior record, mark it needs-review, and re-review; never overwrite historical mappings.

Rules

  • Prefer primary framework publications and official technique catalogs.
  • Keep framework IDs version-qualified; never silently merge revisions.
  • Separate a mapped edge from a control claim, measured efficacy, or compliance assertion.
  • Link shared evidence fields for observation provenance. For mapped observations, carry engagement_scope_ref, skill_version, source, observed_at, finding_status, evidence_refs, limitations, and accountable_owner.

Anti-Patterns

  • No broad equivalence: One keyword does not establish a mapping.
  • No version drift: Verify the current revision before reusing an ID.
  • No compliance leap: Mapping never certifies a control or outcome.
  • No fabricated completeness: Mark unknown edges and gaps.

References

我的收藏