uncertainty-checklist

To prevent AI from generating code with invalid imports and placeholders, enforce a pre-presentation behavioral self-audit to verify import validity, API call accuracy, type correctness, and the absence of TODOs or FIXMEs.

An uncertainty checklist for AI code generation is a systematic self-audit mechanism that checks generated code for import validity, API call accuracy, and type correctness before output, ensuring all code is grounded in verified information and preventing overconfident errors.

Verify tRPC procedure calls and type definitions before presenting generated code by running a pre-output self-audit that confirms API endpoints match scanned definitions and ensures referenced types and fields are correctly understood from definitions.

Yes, you can use a self-audit checklist to calibrate AI confidence and qualify uncertain code statements. The checklist prompts the AI to qualify uncertain statements, avoid absolute claims, and eliminate unimplemented functions before presenting code.

The uncertainty checklist requires no external dependencies or specific environment setup. It is a standalone behavioral system that enforces a self-audit process directly within the AI's code generation workflow to verify completeness and accuracy.