tzai-dataviz

Pass

Audited by Gen Agent Trust Hub on Aug 10, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses a bash script to search for an executable named tzai-image across several local application directories in the user's home folder (e.g., .agents/skills, .claude/skills, .cursor/skills).
  • [COMMAND_EXECUTION]: Once the engine is located, it is executed via bash "$ENGINE" to process data visualization requests.
  • [EXTERNAL_DOWNLOADS]: The skill documentation includes an installation command for the author's own dependency (kedoupi/tzai-image-skill) using the npx skills utility.
  • [PROMPT_INJECTION]: The skill interpolates user-provided text directly into a shell command argument (--prompt "<user subject>"). This represents an indirect prompt injection surface where maliciously crafted user input could attempt to influence the shell environment, although the primary purpose is for the engine to receive the visualization subject.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 10, 2026, 02:50 AM
Security Audit — agent-trust-hub — tzai-dataviz