best-practices-subagent
Pass
Audited by Gen Agent Trust Hub on Aug 26, 2026
Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
- [SAFE]: The skill serves as a safety governance framework. It explicitly defines policies to restrict subagent behavior, including denying dangerous shell commands (such as
rm -rf,systemctl,crontab), enforcing finite retry budgets to prevent resource exhaustion, and implementing strict memory access controls to prevent unauthorized data modification. - [COMMAND_EXECUTION]: The skill references and includes a test fixture for a local shell script (
run.sh) used for linting subagent contracts and validating schemas. This is a standard development utility for maintaining skill integrity. - [SAFE]: The subagent architecture includes deterministic gates and required JSON schema validation for inputs and outputs, which mitigates the risk of indirect prompt injection during agent orchestration.
Audit Metadata