Addy Osmani: Equip your coding agent with tests so it can check its own work
addyosmani · x · 2026-10-05
Google's Addy Osmani shares practical advice for getting high-quality output from coding agents: give them a way to verify their own work, with tests as a core mechanism. His recommendations:
- Write end-to-end tests simulating real user flows as ground truth
- Use property-based testing to define what should never happen and auto-generate thousands of adversarial cases
- When replacing a system, diff old vs new on randomly selected inputs
- Keep tests fast and deterministic — slow or flaky loops teach agents to retry instead of actually fixing bugs
His core distinction: example tests say what should happen; properties say what must not.
Related event: Google Engineer: Equip AI Coding Agents with Tests for Self-Verification(2 posts)→
More from coding & agent
- Photon raises $4.5M seed for open-source messaging layer connecting AI agents to iMessage, WhatsApp, Telegram — SucceededMind · 2026-10-05
- Human-AI collaboration pushes 11-square packing lower bound to 3.875, closing 95.89% of the gap — ctjlewis · 2026-10-05
- Follow-up on the same 11-square packing breakthrough: 300-line verifiable proof via 1,121 weighted dots — ctjlewis · 2026-10-05
- Mobilerun scores perfect 116/116 on Google's AndroidWorld, beating Artemis — SimplyAnnisa · 2026-10-05
- Gemini Robotics-ER 2 plans dual Franka FR3 manipulation, open-sourced with Isaac Sim — Stefania_druga · 2026-10-05
- Dimillian: as models get smarter, intent-based single-thread prompting beats agent orchestration — Dimillian · 2026-10-05