AI coding

Acceptance checks for AI agent tasks

• A Test List That Survives an Agent's Own Edits
• How to Write Acceptance Checks an Agent Can Run Itself
• Turning a Manual QA Checklist into Agent-Runnable Checks
• What an Acceptance Check Costs in Command Time and Tokens
• Writing Acceptance Checks an Agent Can Run Itself
• A Characterization Run in CI: Cost, Time and What to Cache
• A Lint Rule as a Test for Agent Output
• A Test Coverage Budget for Agent Work: When to Stop Adding Tests
• Check Schedule: Always, on Diff, or on Intervals
• Grading Agent Output When There Is No Test to Run
• Replaying Production Traffic Against Agent Changes
• Standard Output and Exit Code Conventions for Agent Checks
• Turning a Bug Report into a Test List an Agent Can Act On
• Turning a Caller Set into Commands: A Test List with a Cost Cap
• What a Check Costs in Command Time and Tokens
• What to Do When an Agent's Check Passes but the Output Is Wrong
• When a Test List Grows Too Long: Pruning Without Losing Coverage
• When to Cache Acceptance Check Results Across Agent Runs
• Where Acceptance Checks Run: CI, Sandbox or Agent Shell
• Where Acceptance Checks Should Run: CI, Sandbox, or Agent Shell
• Writing a Definition of Done an Agent Can Verify