skills/jamdesk/skills/blur-image/Gen Agent Trust Hub

blur-image

Pass

Audited by Gen Agent Trust Hub on Mar 5, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill invokes the magick (ImageMagick) command-line utility to perform image analysis and blurring. It uses structured subcommands such as identify to retrieve dimensions and -region with -gaussian-blur for targeted redaction. These operations are essential for the skill's stated purpose.
  • [EXTERNAL_DOWNLOADS]: The skill provides the user with installation commands and links for ImageMagick. These references target official and well-known repositories, including macOS Homebrew, Ubuntu/Debian official package managers, and the official ImageMagick website.
  • [PROMPT_INJECTION]: The skill processes untrusted image content via multimodal vision to identify redaction regions, which introduces an indirect prompt injection surface.
  • Ingestion points: Image files processed during the identification phase.
  • Boundary markers: No specific delimiters or instructions are used to distinguish image text from system instructions in the vision processing stage.
  • Capability inventory: The agent can execute file system checks and magick subprocesses for image manipulation.
  • Sanitization: While the agent is instructed to focus on coordinate calculation and classification, it does not explicitly sanitize the text content extracted from the image before it is used for internal decision-making.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 5, 2026, 05:05 PM