image-text-replace

Pass

Audited by Gen Agent Trust Hub on Aug 29, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONDATA_EXFILTRATION
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses the meitu CLI tool to perform image processing tasks. It constructs shell commands using parameters derived from user input, including image_url, source_words, and target_words.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied strings to generate prompts for image editing models (e.g., creating a prompt like "Replace the text 'xxx' with 'yyy' in the image").
  • Ingestion points: User-provided source_words, target_words, and prompt parameters processed in SKILL.md.
  • Boundary markers: None explicitly defined in the prompt generation logic to isolate user data from the system command instructions.
  • Capability inventory: Subprocess execution of the meitu binary as defined in the exec permissions.
  • Sanitization: None described; the skill interpolates strings directly into the CLI command template.
  • [DATA_EXFILTRATION]: The skill is configured to read from ~/.meitu/credentials.json. While this is a sensitive file path, it is the standard location for the vendor's own CLI tool credentials and is used to authenticate requests to the vendor's API as part of the skill's primary functionality.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 29, 2026, 02:53 AM
Security Audit — agent-trust-hub — image-text-replace