Building with coding agents
The first prompt was the easy part.
Practical field notes for debugging generated code, choosing the right agent, testing the real flow, and deciding what is safe to ship.
Get the useful bits by email.
New field notes and honest tool comparisons. No product pitch pretending to be a newsletter.
Start here
Debug the app the agent built

Guide · Updated August 2026
How to fix a vibe coded app with a debugging loop
How to fix vibe coded app failures after an agent built them: capture the screen, investigate with the right evidence, verify the original path, and deploy.
Read the guide →Latest field notes
Comparison
Codex vs Claude Code in 2026
Codex vs Claude Code in 2026: capability, workflow, pricing, and which agent to reach for — with real usage, not spec sheets.
Read →Guide
Lovable vibe coding: when your app breaks
Lovable's Try to Fix button is free — until the error keeps coming back. The debugging ladder for Lovable apps, from free retries to a real coding agent.
Read →Guide
React Native debugging, the practical guide
React Native debugging in 2026, the practical guide: DevTools, native logs, the bugs that only show on device, and closing the loop with your agent.
Read →Comparison
Gemini CLI vs Claude Code: quotas, cost, workflow
Gemini CLI vs Claude Code in 2026: account lanes, quotas, permissions, and which agent fits a real terminal workflow. Tested on live projects, not spec sheets.
Read →Guide
Expo testing on device: Expo Go, EAS, TestFlight
Expo testing in 2026: Expo Go vs a dev build, simulators, EAS, TestFlight, and catching bugs in the exact real-device state you already observed on your phone.
Read →Guide
How to show your coding agent a visual bug
"No, the OTHER red box." How to show a coding agent what's visually wrong instead of describing it — screenshots, markup, and visual context that works.
Read →Browse by job
Find the next useful decision.
Guides
Field guides for fixing AI-built apps — vibe debugging, screenshot-to-fix workflows, and mobile QA for solo builders.
Comparisons
Hands-on comparisons of the coding agents people actually build with — workflow, limits, price, and the job each one fits.
Best tools
Tested roundups for people building with AI agents — with the one question other lists skip: does the tool just report the problem, or does it get fixed?