agent-evaluation

Pass

Audited by Gen Agent Trust Hub on Jun 17, 2026

Risk Level: SAFECOMMAND_EXECUTIONREMOTE_CODE_EXECUTIONPROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses the Bash tool to perform code exploration, environment validation, and execution of MLflow CLI commands for tracking and evaluation.- [REMOTE_CODE_EXECUTION]: The skill includes automation to generate and execute Python scripts based on internal templates. This involves dynamically creating new files like run_agent_evaluation.py and executing them to run evaluations.- [PROMPT_INJECTION]: The skill processes external datasets through the mlflow.genai.evaluate API, creating a surface for indirect prompt injection. Ingestion points: datasets processed in mlflow.genai.evaluate (SKILL.md). Boundary markers: Absent. Capability inventory: Bash, Write, WebFetch (SKILL.md). Sanitization: Absent (SKILL.md).- [EXTERNAL_DOWNLOADS]: The skill utilizes WebFetch to access documentation from mlflow.org, which is an official resource for the primary tools used in this skill.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 17, 2026, 08:34 PM
Security Audit — agent-trust-hub — agent-evaluation