AI agents

Schema compliance and testing across models

• A Compliance Scorecard for Two Candidate Models
• A Weekly Rollup of Schema Misses by Field and Prompt Version
• Counting Schema Misses by Field, Model and Prompt Version
• Measuring Schema Compliance per Model and Prompt
• Schema Compliance Across Three Providers: A Repeatable Harness
• Schema Compliance by Model: A Comparison Test You Can Run
• Storing LLM Outputs in Postgres: JSONB and Queries
• Testing Structured Outputs at Scale: Fixtures and Real Calls
• Two Models, One Input: Comparing First-Failing Paths
• When a Schema Miss Is a Prompt Bug, Not a Model Bug