core-data

Pass

Audited by Gen Agent Trust Hub on Apr 23, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill provides standard documentation and code snippets for the Core Data framework. The logic follows official development guidelines for stack initialization, background context usage, and UI integration.
  • [DATA_EXFILTRATION]: The skill contains links to sosumi.ai for external documentation. These are informational references to a documentation mirror and do not involve automated network requests or exfiltration of sensitive information.
  • [PROMPT_INJECTION]: The skill identifies an attack surface for indirect prompt injection via data ingestion points in batchImport and importItems in SKILL.md. Evidence: Ingestion points are the record dictionaries and item data; Boundary markers are absent; Capabilities include database persistence via context.save(); Sanitization is absent in these boilerplate snippets. These are standard implementation examples for a persistence layer with no malicious intent detected.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 23, 2026, 01:38 AM