Matt Pocock: AI test quality depends on codebase design, banning tests is wrong
mattpocockuk · x · 2026-10-08
TypeScript educator Matt Pocock shares his experience with AI agents writing tests: quality depends on the codebase's module design — better codebase, better tests. He advocates testing in larger units and enforcing those boundaries with dependency checkers. Tautological tests are a plague, but banning all tests is the wrong answer.
Related event: Matt Pocock: AI Test Quality Hinges on Codebase Modularity(2 posts)→
More from coding & agent
- Theo: agents wrote 200+ buggy PR-watcher scripts; shares a prompt to audit the waste — 0xkarasy · 2026-10-08
- Agent multiplexer tips: vote on responses and let a sweep task learn from them — jasonkneen · 2026-10-08
- Grok monitors X feedback and auto-files bugs — 'the loop is complete' — RachelVT42 · 2026-10-08
- Moyai, open-source cloud coding agent, cuts Devin bills 79% from $101,872 to $21,700 — tristanbob · 2026-10-08
- Pi coding agent gets an auto-mode plugin that locally classifies risky shell commands — deepu105 · 2026-10-08
- Matt Pocock shares /wait-what prompt to surface an agent's scattered questions — mattpocockuk · 2026-10-08