Agent Verification
Domain: Agent Architectures
🎯 Motivation
After an agent completes a task, verifying the result is correct before accepting it.
📋 Concrete Example
Agent says ‘I fixed the bug.’ Verification: run the tests. Tests pass? Yes -> accept. No -> reject and ask agent to try again. Verification ensures quality.
🔗 Analogy
A teacher grading homework - the student says ‘I am done.’ The teacher checks the work. Correct? Accept. Wrong? Hand back for revision. Agent verification does this for AI outputs.