dby-image

Warn

Audited by Gen Agent Trust Hub on Sep 13, 2026

Risk Level: MEDIUMDYNAMIC_EXECUTIONCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [DYNAMIC_EXECUTION]: The script scripts/lib/locate-dby.mjs implements a dynamic module loader that resolves the path to a sibling package (dby-api) by searching parent directories. It then uses the import() function to load code from a computed file path at runtime. This behavior is identified as dynamic loading from computed paths.
  • [COMMAND_EXECUTION]: The script scripts/wechat-crops.mjs executes the system utility sips via execFileSync to perform image cropping and resampling. Additionally, scripts/dby.mjs uses spawn to execute a shell command to run a script found in the sibling dby-api package.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external Markdown files and user-provided prompts. The script scripts/plan-figures.mjs parses Markdown content to extract text for image prompts, creating a surface where malicious instructions embedded in the processed data could influence agent behavior. Boundary markers are absent in this processing stage, and the skill possesses capabilities such as network access to the vendor API and local file write operations.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Sep 13, 2026, 02:42 AM
Security Audit — agent-trust-hub — dby-image