mf-shared-deps
Warn
Audited by Gen Agent Trust Hub on Mar 6, 2026
Risk Level: MEDIUMCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses unsafe shell interpolation in
SKILL.md. The commandnode scripts/shared-config-check.js --context '<MFContext-JSON>'allows for command injection if the context (derived from project files via themf-contextskill) contains shell-sensitive characters like single quotes or backticks. - [PROMPT_INJECTION]: The skill processes untrusted external data, creating an indirect injection surface.
- Ingestion points: The
mf-contextskill collects configuration and artifact data from the target project files. - Boundary markers: No shell-escaping or delimiters are used in the command execution step to isolate the data from the command.
- Capability inventory: The skill utilizes the
Bash(node *)tool for execution, providing a pathway to the system shell. - Sanitization: No input sanitization or shell-escaping logic is implemented for the JSON input before it is passed to the command line.
Audit Metadata