image-search

Pass

Audited by Gen Agent Trust Hub on Jun 22, 2026

Risk Level: SAFECOMMAND_EXECUTIONCREDENTIALS_UNSAFEPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes the meitu binary through the shell to perform image searches.
  • [CREDENTIALS_UNSAFE]: The skill is designed to read sensitive API keys (MEITU_OPENAPI_ACCESS_KEY, MEITU_OPENAPI_SECRET_KEY) from local files (~/.meitu/credentials.json) and environment variables. This access is consistent with the standard authentication workflow for the vendor's (meitu) own tool.
  • [PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface through data ingestion and command interpolation.
  • Ingestion points: User-provided prompt keywords and image_ids enter the agent context via the image_search tool definition.
  • Boundary markers: No explicit delimiters or warnings to ignore embedded instructions are present in the command interpolation logic.
  • Capability inventory: The skill has the capability to execute the meitu subprocess and write files to ./output/ and ~/.openclaw/workspace/visual/.
  • Sanitization: There is no evidence of sanitization or escaping of the user-provided prompt array before it is passed to the shell command.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 22, 2026, 04:39 AM
Security Audit — agent-trust-hub — image-search