label-capture-web

Pass

Audited by Gen Agent Trust Hub on Sep 14, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill references and directs the agent to fetch content from official vendor-owned domains and well-known services, which is standard for developer documentation tools.
  • Fetches API references and guides from docs.scandit.com and ssl.scandit.com.
  • References official sample code from the vendor's GitHub organization (github.com/Scandit).
  • Directs the use of the cdn.jsdelivr.net service for hosting SDK library files.
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests user input and external documentation to generate code, creating a surface for indirect prompt injection.
  • Ingestion points: Data entered by the user (label descriptions) and external documentation fetched via URL in SKILL.md.
  • Boundary markers: The instructions explicitly command the agent to verify all code against the provided references and avoid relying on internal training data.
  • Capability inventory: The skill allows the agent to generate integration code and instructs the agent to write it to files and execute setup commands like npm install.
  • Sanitization: The skill relies on the underlying model's code generation guardrails; no explicit sanitization logic is provided in the instructions.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 14, 2026, 04:32 PM
Security Audit — agent-trust-hub — label-capture-web