before-after

Pass

Audited by Gen Agent Trust Hub on Aug 10, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: Reads and writes project-specific data and markdown files to the local filesystem, defaulting to the ~/jtbd directory. This behavior is essential for its primary function of managing and persisting transformation data bundles.
  • [EXTERNAL_DOWNLOADS]: Integrates with external image generation services such as the ChatGPT API or /nano-banana to create visual infographics based on final grid data.
  • [PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface due to its ingestion of external JSON data from project files. • Ingestion points: Processes content from jtbd.json during the dimension mapping phase. • Boundary markers: None specified to delimit input data from instructional prompts. • Capability inventory: Performs file system writes and initiates network requests for image generation. • Sanitization: Does not explicitly validate or filter external JSON content before interpolation.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 10, 2026, 09:32 PM
Security Audit — agent-trust-hub — before-after