agent-studio-evaluations
Pass
Audited by Gen Agent Trust Hub on Nov 6, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill frequently executes the
algobotCLI using thesubprocessmodule in Python. These commands are used for legitimate administrative tasks such as fetching evaluation metrics, creating experiments, and patching agent configurations. - [EXTERNAL_DOWNLOADS]: The skill requires the installation of the
algobot-aipackage from NPM. This is an official tool provided by the skill's author (Algolia) and is used for interacting with their Agent Studio platform. - [INDIRECT_PROMPT_INJECTION]: The skill includes functionality to import real user conversation history for evaluation purposes. To mitigate the risk of processing sensitive or malicious data embedded in these histories, the skill implements a PII redaction layer in
scripts/build_history_dataset.pythat filters emails, phone numbers, and credit card numbers using validated patterns. - [SAFE]: The skill is a standard developer toolset authored by a trusted vendor. It follows secure practices for handling sensitive data and managing external command execution without using dangerous patterns like shell interpolation or obfuscation.
Audit Metadata