The Agent Skills Directory

[COMMAND_EXECUTION]: The skill defines a core execution workflow that uses the meitu CLI tool. The command construction pattern meitu image-poster-generate --prompt "<desc>" directly interpolates the user-provided <desc> into a shell command string. This is a high-risk command injection vulnerability as an attacker could provide a prompt containing shell metacharacters (e.g., ;, &, |, `) to execute arbitrary commands on the host system.
[CREDENTIALS_UNSAFE]: The skill is configured to read sensitive API keys, specifically MEITU_OPENAPI_ACCESS_KEY and MEITU_OPENAPI_SECRET_KEY, from the file ~/.meitu/credentials.json. Direct access to plaintext credential files in the home directory is a significant security risk.
[DATA_EXFILTRATION]: The skill requests and uses permissions to read several sensitive files, including ~/.meitu/credentials.json and ~/.meitu/tool-registry.json. This provides the skill access to vendor-specific authentication data and system configuration information.
[PROMPT_INJECTION]: While the skill does not contain malicious prompt injection patterns itself, it creates a surface for indirect injection by processing user-supplied descriptions and reference image URLs. These inputs are used as parameters for the image generation tool without evidence of sanitization or validation, which, combined with the shell execution pattern, escalates the risk.

image-poster-generate