Data Privacy & Governance

"Govern data like a liability until proven an asset: know what it is, where it flows, and who may touch it." [INFERENCE]

TL;DR

Applies data privacy governance: privacy-by-design (PII/PHI classification, minimization, encryption, anonymization/pseudonymization), retention/access policy, and tamper-evident audit evidence. It does not own data-quality pipelines or catalog/lineage strategy. [EXPLICIT][CONFIG]

For Claude Platform this skill is supporting, never the primary API owner: it reviews PHI/schema-cache boundaries, ZDR/retention claims, file privacy, computer/memory evidence, and cache diagnostics after the relevant runtime/server/client/files/context skill has established the API contract. [CONFIG]

When to use

• Applying privacy-by-design to a data flow (classify PII/PHI, minimize, encrypt, anonymize). [EXPLICIT]
• Designing audit trails (tamper-evident, temporal, compliance-ready). [EXPLICIT]
• Reviewing privacy boundaries around a data pipeline; route the quality implementation to data-quality. [CONFIG]
• Reviewing classification/retention implications of a catalog; route catalog and lineage strategy to data-strategy. [CONFIG]

Sub-capabilities (resource map)

Procedure

1. Classify data (public/internal/PII/PHI); minimize collection to purpose. [DOC]
2. Protect: encrypt at rest/in transit; anonymize/pseudonymize where identity is not needed. [DOC]
3. Audit: tamper-evident, append-only trails with who/what/when; align to retention policy. [DOC]
4. Route schema/quality implementation to data-quality and catalog/lineage ownership to data-strategy; retain only privacy requirements and evidence here. [CONFIG]

Quality Criteria

• Data classified; collection minimized to purpose. [DOC]
• PII/PHI encrypted; anonymized where identity is unneeded. [DOC]
• Audit trails tamper-evident with retention aligned to policy. [DOC]
• Quality and catalog work has an explicit handoff rather than duplicate ownership. [CONFIG]
• Claims evidence-tagged. [EXPLICIT]

Anti-Patterns

• Collecting PII "just in case" with no purpose or retention limit. [DOC]
• Mutable/over-writable audit logs (non-evidentiary). [DOC]
• Claiming ownership of generic quality, observability, or catalog implementation from a privacy review. [CONFIG]

Contract

• Aceptación: datos clasificados y minimizados; PII/PHI cifrada/anonimizada; retención/acceso definidos; audit trails tamper-evident; handoffs explícitos para calidad y catálogo. [EXPLICIT][CONFIG]
• Límites: capa de privacidad; no implementa calidad (data-quality), catálogo/lineage/estrategia (data-strategy) ni modela esquemas (database-design). [EXPLICIT]
• Casos borde: anonimización reversible vs irreversible; derecho al olvido vs trails inmutables → diseñar conciliación. [INFERENCE]
• Supuestos: marco regulatorio aplicable identificable (GDPR/CCPA/HIPAA según sector). [SUPUESTO]
• Trade-off: controles de privacidad reducen riesgo legal a cambio de fricción operativa y menos datos crudos. [EXPLICIT]

Related Skills

• data-strategy — sets direction this governs. [EXPLICIT]
• data-quality — owns pipeline validation and quality gates. [CONFIG]
• database-design — schemas this classifies and protects. [EXPLICIT]
• application-security — encryption/access controls at the app layer. [EXPLICIT]
• claude-api-server-tools, claude-api-client-tools, claude-api-files-documents, claude-api-context-management — own the API mechanics this skill reviews for privacy and retention. [CONFIG]

Packet

Capas del packet, cargables bajo demanda (disciplina ICM: una capa por vez, nunca todas juntas): references/ guías de profundidad (cargar UNA por etapa) · knowledge/ cuerpo de conocimiento · prompts/ prompts listos · examples/ salida de ejemplo · agents/ subagentes del packet · assets/ recursos estáticos.