youmind-blog-cover

Pass

Audited by Gen Agent Trust Hub on Mar 31, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill installs the @youmind-ai/cli package from the NPM registry. This is a vendor-managed tool required for the skill to communicate with the YouMind service.\n- [COMMAND_EXECUTION]: Shell commands are used to invoke the youmind CLI for board discovery, chat creation, and polling. A local script, scripts/extract-images.js, is also executed to parse the results.\n- [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface by directly interpolating user-provided article titles or URLs into instructions for the image generation model.\n
  • Ingestion points: User-provided topics, titles, or URLs processed in the Step 3 workflow of SKILL.md.\n
  • Boundary markers: No delimiters or "ignore instructions" warnings are used when inserting user content into the prompt template.\n
  • Capability inventory: The agent has access to shell commands (vendor CLI) to transmit the resulting prompt.\n
  • Sanitization: No validation or escaping is applied to user input before it is used to construct the AI prompt.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 31, 2026, 03:39 AM