riansanderson.bsky.social

@riansanderson.bsky.social

digital plumber. figuring out how these masses of agentic minions can help us write higher quality software.

7 months ago I read about gas town and today I read the new thing. what worries me now is either he’s gotten less insane or I’ve gotten more insane.

Git Archaeology and Project History tasks are a fantastic thing to hand off to LLM harnesses. Dev who hasn't worked on this specific app for years was sure they had added support for XYZ. Claude produced a detailed summary of when XYZ was added and later removed, with commit SHAs and context.

Even before the agentic revolution, prompting tricks stopped being very valuable, as our research has shown. The format isn't key, the best approach to AI right now is to clearly specify your goals, your output, what "good" & bad look like, how to test the results... (yes, this is just management)

BildBildBildBild

LLMs are not speed booster for (serious) coding. They are quality booster if used correctly. LLMs also allow fast prototyping, but that is a completely different story. Relevant in this context -> simonwillison.net/2026/May/31/... @simonwillison.net has a positive take on this.

The solution might be cancelling my AI subscription

I find this post by David Wilson very relatable. David lists 16+ projects he's spun up with AI tooling, and concludes: I didn't mean to build most of these things. …

simonwillison.net

When you create a new empty React app, how many people have the ability to run code on your computer? Let's say you get started with, bun create vite my-app --template=react-ts Unique deps: 180 Unique maintainers: 168 168 targets for a supply chain attack ... just to start.

AI exists and traditional software engineering interviews were already absurd. I propose that interviews now consist of playing a minimum of 4 hours of Factorio together to determine if you'd want to work with someone.