graph-search
Pass
Audited by Gen Agent Trust Hub on Apr 13, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The instruction for running the query script (
python3 ... --query "<term>") directly interpolates the user-provided search term into a bash command. This presents a command injection risk if the user input includes shell metacharacters like backticks, semicolons, or dollar signs. - [PROMPT_INJECTION]: The skill has an indirect prompt injection surface as it processes user input that controls the parameters of a command execution.
- Ingestion points: User-provided search term captured in the
<term>placeholder inSKILL.md. - Boundary markers: Absent; the skill does not instruct the agent to use delimiters or ignore instructions within the input.
- Capability inventory: Subprocess execution via a bash call to a Python script.
- Sanitization: Absent; there are no instructions for the agent to escape or sanitize the input before command interpolation.
Audit Metadata