The Coverage Loop: Turning Free AI Tokens into Verified C++ Tests

Description and use

A feedback loop with AI-generated tests achieves high coverage for C++ code.

Problem

Generated tests from free AI models often lack assertions and coverage, leading to ineffective unit tests.

Outcome

The loop reached 90% line coverage for the string utility by iteratively refining AI-generated tests based on coverage feedback.

Workflow

• Send header to AI model asking for tests
• Compile and run the generated tests
• Run gcov to measure line coverage
• Extract missing lines from gcov output
• Feed missing line hints back to the AI and repeat until coverage plateaus

Evidence

Five steps. Each one is small.

Source

The target: a small C++ string utility. The goal: reach 90% line coverage with generated tests.

Source