image-outfit-swap

Pass

Audited by Gen Agent Trust Hub on Aug 29, 2026

Risk Level: SAFECREDENTIALS_UNSAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [CREDENTIALS_UNSAFE]: The skill is configured to read sensitive credential files from ~/.meitu/credentials.json. While these are vendor-specific credentials required for the Meitu OpenAPI, direct access to credential storage files on the filesystem is a high-sensitivity operation.
  • [COMMAND_EXECUTION]: The skill executes the meitu CLI tool to process images, which involves running shell commands with parameters derived from user input.
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted user input via the prompt parameter and interpolates it into a shell command (meitu image-outfit-swap ... --prompt "<target outfit>"). This pattern presents a vulnerability surface for command injection.
  • Ingestion points: The user-supplied prompt and image_url strings in the Execute section of SKILL.md.
  • Boundary markers: The command template uses double quotes to wrap the prompt, but there are no explicit instructions or mechanisms defined within the skill to sanitize or escape shell-metacharacters.
  • Capability inventory: The skill has permissions to execute the meitu binary and write output files to the local filesystem (./output/).
  • Sanitization: No input validation, character escaping, or sanitization logic is specified for the external content before it is processed by the shell.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 29, 2026, 02:53 AM
Security Audit — agent-trust-hub — image-outfit-swap