ae-experiment-insight
Pass
Audited by Gen Agent Trust Hub on Aug 14, 2026
Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill utilizes a provided Python script
scripts/analyze_experiment.pyto perform deterministic statistical calculations (SRM, t-tests, z-tests). This is a security-positive practice that offloads mathematical computation from the LLM to a controlled, local environment. - [COMMAND_EXECUTION]: The skill interacts with the experiment platform exclusively through a dedicated CLI tool,
ae-cli. Instructions provide clear guidance on searching, inspecting, and running capabilities using structured JSON inputs, minimizing the risk of arbitrary command construction. - [SAFE]: The skill implements a robust diagnostic workflow that requires the agent to validate data reliability (SRM monitoring, duration sufficiency, novelty effects) before making rollout recommendations.
- [SAFE]: Lifecycle operations (e.g., rolling out or deleting experiments) are explicitly restricted. The agent is instructed to only perform these actions upon separate, explicit user request, preventing autonomous destructive actions during interpretation tasks.
Audit Metadata