Leadership

Graders and checklists

• Grading With a Rubric vs a Reference: Which Catches More
• Routing Between a Cheap First Pass and an Expensive Second Look
• When to Let an Agent Mark Its Own Checklist
• Writing the Grader Prompt for a Second-Model Check
• A Prompt Injection Test Suite for Inbox and Ticket Agents
• Building a Labeled Set When No One Has Time to Label
• Calibrating a Confidence Threshold: From Scores to Decisions
• Choosing a Second Model: Cost, Latency, Independence for Graders
• Grading a Trajectory, Not an Output
• Grading an Agent's Work with a Second Model
• Measuring Whether Your Checkers Catch Errors
• Routing Agent Work by Risk: A Cost and Delay Model
• Structured Output as a Checklist Contract
• Structured Verdicts: A JSON Schema for Pass, Fail and Reason
• The Cost of a False Pass: Estimating What a Missed Error Costs
• What a Grader Should See: Slicing the Run Record for the Checker
• What an Agent Can Check Without a Second Model
• What the Grader Sees: Passing the Run Record
• What to Do When an Agent Fails Its Own Checklist
• When the Grader Is Wrong: Reviewing the Reviewer's Verdicts
• When Two Graders Disagree: Triage Rule for Verdicts
• Writing the Reference Answer: One Example, a Small Set, or a Spec