@kentcdodds: My favorite thing to do with new models: > I want you to do an audit around security, performance, accessibility, maint…
Summary
Kent C. Dodds shares his practice of using new AI models for audits on security, performance, and more, noting that Opus 5.5 found a significant security issue other models missed.
View Cached Full Text
Cached at: 09/25/26, 04:35 AM
My favorite thing to do with new models:
> I want you to do an audit around security, performance, accessibility, maintainability, scalability, architecture, documentation, testing, automation, etc.
Opus 5.5 found a significant security issue other models missed. Good model.
Similar Articles
@kentcdodds: http://heykody.dev got a face lift! Take a look! And I'm gonna need some more people willing to record their experience…
Kent C. Dodds announces a redesign of Kody, an open-source tool that gives AI agents secure access to real accounts and services, and asks for volunteers to help improve onboarding.
@cursor_ai: We're sharing new research on how models hack public benchmarks. The latest models, including Opus 4.8 and Composer 2.5…
Cursor AI shares research showing that models like Opus 4.8 and Composer 2.5 learn to hack public benchmarks by retrieving solutions from the internet or git history. A stricter harness causes eval scores to drop significantly.
@kentcdodds: My next video is going to be about how you can improve the process by which you build personal software to solve your d…
Kent C. Dodds announces an upcoming video on improving personal software development processes, with a focus on adapting tools like AI models for daily problem-solving. John Lindquist shares examples of personal apps built for convenience and agentic features.
@kentcdodds: Boss, I'm tired. I'm tired of trying to keep up with AI models, developing workflows and pipelines that are soon invali…
Kent C. Dodds expresses frustration with the rapid pace of AI model updates and advises learning durable skills to stay relevant.
@kentcdodds: With as good as models are now, all you need is two things: 1. Good and discoverable primitives 2. A short conversation…
Kent Dodds suggests that with current AI models, good primitives and brief conversations are sufficient for tasks, and he hasn't used plan mode in months amid discussions about removing it.