arc42-section-05

Pass

Audited by Gen Agent Trust Hub on Jun 22, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The generated PlantUML code references standard C4-PlantUML components hosted on the official plantuml-stdlib GitHub repository (raw.githubusercontent.com/plantuml-stdlib/C4-PlantUML).\n- [PROMPT_INJECTION]: The skill identifying an attack surface for indirect prompt injection as it ingests and processes content from other project documentation sections (3, 4, 6, 7) and user input to generate its outputs. This risk is inherent to its primary function of synthesizing project information.\n
  • Ingestion points: Retrieves data from external project sections (interfaces, strategy, runtime, deployment) and user responses within SKILL.md.\n
  • Boundary markers: No specific delimiters or safety instructions are defined to separate retrieved content from the agent's instructions.\n
  • Capability inventory: The skill is capable of writing generated diagrams to the filesystem in the docs/diagrams/ directory.\n
  • Sanitization: No sanitization logic is implemented for the data retrieved from external project sections.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 22, 2026, 05:08 PM
Security Audit — agent-trust-hub — arc42-section-05