recipe-verify-architecture-diagrams

Pass

Audited by Gen Agent Trust Hub on Jul 11, 2026

Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes the plantuml command line utility to regenerate PNG images from .puml files. It also uses a local Python runner (.plan/execute-script.py) to invoke internal tools for file discovery and configuration management.
  • [PROMPT_INJECTION]: The skill analyzes local source code (.java) and documentation (.adoc, .md) to verify diagram correctness. While this processes untrusted data from the repository which could theoretically contain indirect prompt injection, the risk is minimal as the skill performs structural comparisons and is scoped to specific documentation tasks.
  • Ingestion points: Reads .puml, .adoc, .md, and .java files from the project directory.
  • Boundary markers: No explicit markers are used, but the skill operates within a controlled workflow mode.
  • Capability inventory: File read/edit access and local execution of plantuml and internal scripts.
  • Sanitization: Standard file reading and string comparison; no specific prompt-level sanitization is described.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 11, 2026, 07:13 AM
Security Audit — agent-trust-hub — recipe-verify-architecture-diagrams