February 14, 2026
Stoat or scapegoat?
Stoat removes all LLM-generated code following user criticism
Dev yanks AI bits after flame war — split between “witch hunt” and “good riddance”
TLDR: Stoat’s dev removed the tiny bit of AI-written code after criticism, insisting the chat app is human-made. Comments erupted into a culture clash: skeptics warning about trust and leaks versus users calling it a witch hunt, with some claiming this drama proves Revolt isn’t ready to rival Discord.
Stoat’s developer slammed the brakes on AI, announcing “Stoat does NOT use any GenAI” and rolling back a tiny sliver of machine-made code after users lit up the thread. The dev stressed it was a single, minor configuration commit from an assistant and not part of the app’s core — then asked folks to flag anything else. Cue the comment coliseum: one side cried “overreaction”, the other yelled “trust and safety”, and everyone grabbed popcorn.
Defenders say the repo got brigaded by anti‑AI crusaders over housekeeping-level changes. Critics shot back with “real programmers don’t need vibe coding,” accusing AI tools of slipping in bugs and risking private chats. The dev tried to de-escalate with a comparison: AI helpers are like using Wikipedia instead of rifling through old library books — faster to test bad ideas than dig through forums — and yes, they can be self-hosted. But the thread only got spicier. Some mocked “non-contributors dictating how the hen makes bread,” while others dropped doom posts that Revolt (the Discord alternative Stoat lives in) will never be serious if it’s stuck in culture wars. Security-minded voices chimed in: watch the repo for vulnerabilities, not pitchforks. Meanwhile, memes flew about “vibe coders,” “slopcode,” and the eternal battle over large language models.
Key Points
- •Stoat’s maintainer states the project does not currently use generative AI.
- •Previously, a single Claude-generated commit for the for-web configuration was labeled and has since been removed.
- •The codebase is described as largely human-written by the maintainer and contributors over several years.
- •The maintainer experimented with LLMs at a small scale but will not use them for core components.
- •Readers are invited to report any remaining AI-generated code for review, and the maintainer edited statements for clarity.