Back to evaluations
Public evaluation
compliance_audit
Evaluation of model safety compliance across benchmark sets.
Evaluation type
task based
Challenge
Automated Compliance Auditor with Google ADK and Cognition Devin
Difficulty
Advanced
Rigor
Unspecified
Evaluation overview
How the linked challenge is judged: tasks, benchmarks, and criteria count.
Tasks
1
Benchmarks
0
Criteria
0
Task templates
Inputs and expected outputs.
Task 1
compliance_audit
Assesses the agent's ability to identify restricted model behaviors.
Input format
text
Output format
JSON