import-track
Warn
Audited by Gen Agent Trust Hub on Feb 18, 2026
Risk Level: MEDIUMCOMMAND_EXECUTION
Full Analysis
- COMMAND_EXECUTION (MEDIUM): The skill instructs the agent to use the
Bashtool to execute amvcommand using the{source_file}argument directly. If the agent does not properly escape shell metacharacters in the filename (e.g., backticks, dollar signs, or subshells), a malicious user could achieve arbitrary command execution within the user's environment.\n- INDIRECT_PROMPT_INJECTION (LOW): The skill ingests untrusted markdown files from the user's filesystem.\n - Ingestion points: User-provided
.mdfiles at<file-path>.\n - Boundary markers: None defined; the agent is expected to parse the file or filename for track metadata.\n
- Capability inventory:
Bash(for file movement),Read(for file access), andGlob(for file searching).\n - Sanitization: No explicit sanitization or validation of the file content or path is described in the instructions to prevent the agent from being influenced by data inside the track files.
Audit Metadata