swift-testing

Pass

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: SAFE
Full Analysis
  • PROMPT_INJECTION (SAFE): No evidence of instructions designed to bypass safety filters or override agent behavior. The content is purely educational and instructional.\n- EXTERNAL_DOWNLOADS (SAFE): References to external libraries (e.g., swift-snapshot-testing) are for documentation purposes and represent industry-standard tools. No automated remote code execution or untrusted downloads are performed.\n- DATA_EXFILTRATION (SAFE): No patterns of data collection or exfiltration. Placeholder values in code snippets are used appropriately for demonstration.\n- OBFUSCATION (SAFE): No use of encoding, zero-width characters, or homoglyphs to hide malicious intent.\n- INDIRECT_PROMPT_INJECTION (LOW): The skill ingests user-provided Swift code for testing guidance. While it lacks explicit boundary markers, the skill does not possess high-risk capabilities (like automated shell execution) that could be exploited via this vector.\n
  • Ingestion points: User code snippets provided for testing advice.\n
  • Boundary markers: Not explicitly defined in the skill body.\n
  • Capability inventory: None; the skill is documentation-based and does not execute code.\n
  • Sanitization: Not applicable as the skill only generates text/code samples.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 17, 2026, 05:46 PM