The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The skill is configured to read sensitive credential files located at ~/.meitu/credentials.json and ~/.meitu/tool-registry.json. While these are vendor-owned resources, they contain high-value authentication information.
[COMMAND_EXECUTION]: The skill performs command execution via the meitu binary to process video content.
[REMOTE_CODE_EXECUTION]: The skill constructs shell commands by interpolating user-supplied input into the prompt and video_url parameters: meitu video-element-remove --video_url {url} --prompt "{画面描述}". The lack of sanitization for the {画面描述} (prompt) field allows for command injection; a malicious input containing shell metacharacters (e.g., ;, `, $()) could execute arbitrary code in the host environment.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it processes untrusted data which directly influences system-level execution.
Ingestion points: User-provided video_url and prompt described in the Execute section.
Boundary markers: None; the input is wrapped in double quotes in the shell command but lacks explicit delimiters or instructions to ignore embedded commands.
Capability inventory: Execution of the meitu CLI utility with file-system write permissions.
Sanitization: No evidence of escaping or validation of user-provided strings before they are executed in the shell.

video-element-remove