August 3, 2026
Fork around and find out
Devtools must be open source
Open your coding tools or brace for a community meltdown over who fixes the mess
TLDR: The article argues coding tools should be open source because AI now makes it much easier for people to personalize software and keep those changes updated. Commenters loved the freedom in theory but fought hard over the practical part: if AI-made tweaks break your tools, is that empowerment or just a new headache?
A bold new manifesto on exe.dev declared that developer tools should be open source so people can change them however they want, especially now that AI helpers can supposedly rewrite software for you and even keep your custom changes updated over time. In plain English: instead of begging a company for a feature, you could tell an AI to tweak the app yourself. Sounds empowering, right? The comments immediately turned into a mini food fight.
One camp was basically: yes, this is the dream. Simon Willison vibes were strong in the thread, with people arguing that open source has always mattered because even if you don’t personally edit the code, someone else can. That freedom still counts. But the louder drama came from the skeptics, who were not buying the “just let AI rewrite the tool” fantasy so easily. One commenter practically slammed the brakes: if you customize something and it explodes, who exactly is supposed to fix it? Another called the idea of changing simple settings by rewriting the whole app with an AI wasteful and absurd, like using a bulldozer to hang a picture frame.
And then came the messy product gripes. One paying user said they clicked through to the author’s side project, meat.dev, only to find sparse docs and limited built-in support, which added a delicious layer of irony to a post preaching openness and flexibility. The mood was classic internet: half inspired, half annoyed, and fully ready to argue about it.
Key Points
- •The article says many engineers historically relied on software built by others and often found personal software projects hard to justify because of time and maintenance costs.
- •Crawshaw argues that AI agents now make software personalization easier by handling source checkout, local builds, modifications, and maintenance tasks.
- •He outlines two key workflow prompts: one to modify and document a local source-based customization, and another to automatically fetch upstream changes and rebase local patches on a schedule.
- •The article states that these instructions can be embedded into an open-source agent as discoverable skills, reducing the need for manual setup or traditional programming.
- •Crawshaw presents Shelley as an example of built-in self-customization and introduces meat.dev as a side project focused on reviewing code produced by agents before deployment.