matrixscan-pick-kmp

Pass

Audited by Gen Agent Trust Hub on Sep 2, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest user requirements (such as symbology lists and product data structures) to generate and apply code to the local project filesystem. While this represents a functional capability of the agent, it creates a surface for indirect instructions.
  • Ingestion points: User-supplied symbology types and product identifiers processed in SKILL.md and references/integration.md.
  • Boundary markers: None explicitly defined for interpolating user-supplied strings into code templates.
  • Capability inventory: The skill instructs the agent to apply generated Kotlin and Swift code to local project files.
  • Sanitization: Standard for development assistance; no specific sanitization of user-supplied code snippets is mentioned.
  • [EXTERNAL_DOWNLOADS]: The skill references external dependencies from well-known and vendor-owned sources.
  • Evidence: Directs the agent to Maven artifacts on central.sonatype.com and Swift packages from Scandit's official repositories. These are standard practices for third-party SDK integration and originate from the vendor and established package registries.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 2, 2026, 07:18 AM
Security Audit — agent-trust-hub — matrixscan-pick-kmp