Debug Animated UI Bugs by Having Codex or Claude Code Watch a Screen Recording
Melvin Vivas · X video post · 2026-08-30 · 0:39 · 191 views · Open on X
Topics: AI Dev Tools & Productivity, Evaluation (Evals) & Testing · Level: beginner
Summary
This is a short QA tip for AI coding assistants. Some UI bugs involve animation and won't show up in a screenshot. In those cases, record a video of your manual test until the bug appears, then ask the coding agent to watch it and describe the bug. The creator first showed this with Claude Code. Here he confirms that OpenAI Codex, running a model called "Luna", also found the bug from the video, with less detail but good enough. His point is that you may not need the smartest model for this.
Key points
- Screenshots can't capture bugs that involve animation or transitions, so a video recording works better.
- Workflow: test the feature manually, record your screen until the bug happens, then give the video to the coding agent and ask it what the bug is.
- Claude Code was shown doing this first, in the quoted post.
- OpenAI Codex also found the issue from the video, though it described it in less detail than Claude Code.
- The creator ran Codex with 'Luna', which suggests a smaller or cheaper model can handle this kind of video-based bug finding.
- Practical lesson: pick the model that fits the task. You don't always need the top model for visual QA triage.
Resources mentioned
- OpenAI Codex · tool · openai.com · paid · recommended by both Bashiri Smith & Melvin Vivas
OpenAI's coding agent. In the diagram it writes code, fixes review findings and drives the build loop. The creator also used it to make this video.
Also in: Developers moving from ChatGPT/Codex to Claude (Melvin Vivas on X · notes), Creator Opinion: Codex Is Still the Best at Computer Use (Melvin Vivas on X · notes), OpenAI Codex Ambassador applications are open (Melvin Vivas on X · notes), GPT-Live-1.1: Multilingual Voice Update for ChatGPT and Codex (Melvin Vivas on X · notes) and 264 more - Claude Code · tool · code.claude.com · paid · recommended by both Bashiri Smith & Melvin Vivas
Build agents and pipelines from the terminal; the guide's main agentic coding tool.
Also in: First impressions: Claude Code in the cloud is good and fast (Melvin Vivas on X · notes), Why Claude Code feels faster: gzipped turns and instant cloud sessions (Melvin Vivas on X · notes), Watch What Fills Your Coding Agent's Context and When to Clear (Melvin Vivas on X · notes), Claude Code's Right Sidebar and Context Breakdown (Melvin Vivas on X · notes) and 124 more - Luna max · tool · developers.openai.com · paid
The model the creator ran in Codex for this test; the post presents it as not the smartest option.
Also in: Watching Codex weekly limits: GPT 5.6 Sol vs Luna max cost (Melvin Vivas on X · notes)
Try this
- When a bug involves animation, record a video of your manual test until the bug shows up instead of taking a screenshot.
- Give the recording to Claude Code or Codex and ask it to watch the video and explain the bug.
- Try a smaller or cheaper model first for video-based bug finding before moving to a stronger one.
More in AI Dev Tools & Productivity
- Plan With a Frontier Model, Code With a Local Quantized Qwen
- Synthetic (synthetic.new): A Clear Way to Show LLM Subscription Usage Limits
- herdr: Managing Codex and Claude Agents Across Projects
- Claude Code Tip: Debug Animation Bugs From a Screen Recording
- Codex Tip: Ask Codex to Organize Your Recent Chats
- Codex Luna reserve has its own separate weekly limit