skills/christowles/blog/blog-image/Gen Agent Trust Hub

blog-image

Pass

Audited by Gen Agent Trust Hub on Apr 20, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection as it ingests and processes untrusted markdown files to generate image prompts without explicit boundary markers or instructions to ignore embedded commands.
  • Ingestion points: Markdown files specified via $ARGUMENTS or detected through git status.
  • Boundary markers: Absent.
  • Capability inventory: Executes shell commands (pnpm tsx) and writes files to the local filesystem.
  • Sanitization: None described; the generated prompt is directly interpolated into a shell command.
  • [COMMAND_EXECUTION]: The workflow involves executing a shell command (pnpm tsx scripts/generate-blog-image.ts "<prompt>" ...) where the <prompt> variable is derived from the analysis of external blog content. This creates a command injection surface if the generated prompt contains shell-sensitive characters (like double quotes, backticks, or semicolons) and the agent does not properly escape them before execution.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 20, 2026, 12:54 PM
Security Audit — agent-trust-hub — blog-image