Back to all materials

AI for Builders · Workshop 08

Testing AI-Generated Code

Verify code produced or modified by AI before you trust it. The skill this workshop builds is proof: showing that what the AI generated works, matches the original request, avoids regressions, and is safe to merge. You will turn requirements into acceptance criteria, hunt edge cases, combine unit tests, manual testing, and review, and leave with a workflow you can reuse on real changes.

By Martin-Patrick Larouche August 14, 2025

Duration
45 to 60 minutes
Level
Beginner to Intermediate
Tools
Claude or another AI assistant, a code editor, an optional test framework, and an optional Git repository
Outcome
A repeatable workflow for testing AI-generated code before accepting it

You will leave with

  • An Acceptance Criteria Template you fill in before testing
  • An AI-Generated Code Testing Checklist for the whole validation pass
  • An Edge Case Discovery Prompt for the scenarios the happy path forgets
  • A Manual Testing Plan for the things a unit test cannot prove
  • A Test Review Checklist for judging the tests themselves
  • A Safe Merge Checklist for the final accept-or-reject decision

References

  1. ↑CodeRabbit, "State of AI vs Human Code Generation Report" (2025)
  2. ↑H. Huang et al., "More Code, Less Reuse: Investigating Code Quality and Reviewer Sentiment towards AI-generated Pull Requests" (arXiv, 2026)
  3. ↑J. Spracklen et al., "We Have a Package for You! A Comprehensive Analysis of Package Hallucinations by Code Generating LLMs" (arXiv, 2024)
  4. ↑Aikido Security, "Slopsquatting: The AI Package Hallucination Attack Already Happening" (2026)