agent-tester
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill defines lifecycle hooks (
hooks.preandhooks.post) that execute shell commands, includingnpm testandjq, to automate test suite execution and parse results. - [INDIRECT_PROMPT_INJECTION]: The skill possesses an indirect prompt injection vulnerability surface typical of code execution and testing tools:
- Ingestion points: Implements workflows that process the
$TASKvariable and run test files within the local environment. - Boundary markers: Lacks specific delimiters or guardrails to prevent untrusted test code or inputs from influencing agent behavior.
- Capability inventory: Retains shell command execution capabilities via lifecycle hooks.
- Sanitization: Does not enforce sanitization or filtering on the code or test outputs being processed.
Audit Metadata