heuristic
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: Step 6 instructs the agent to execute
node scripts/eval.mjs <path> --json. The<path>parameter is derived from the user-provided$ARGUMENTS. If the platform does not properly sanitize these arguments before shell execution, it creates a potential command injection surface. - [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze external UI content, which provides an attack surface for instructions embedded in the audited data to influence agent behavior.
- Ingestion points: UI source files located at the path specified in
$ARGUMENTS. - Boundary markers: Absent; the instructions do not specify the use of delimiters or 'ignore' instructions for the data being audited.
- Capability inventory: The skill possesses the ability to execute subprocesses (via
node) and read local file systems. - Sanitization: Absent; there are no instructions to sanitize, validate, or filter the UI content before processing.
Audit Metadata