agent-evaluation

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.90). The skill's Documentation Access Protocol explicitly requires querying and fetching https://mlflow.org/docs/latest/llms.txt at runtime (and using WebFetch on any documents it references), and those fetched documents are used to form prompts/instructions for dataset creation, scorers, and evaluation—i.e., remote content directly controls agent behavior and is a required runtime dependency.

MEDIUM W021: Hidden or invisible Unicode characters detected (potential obfuscation or prompt injection).

• Hidden Unicode characters detected (1 type(s) found)