October 28, 2025
When vibes meet version control
Your vibe coded slop PR is not welcome
AI code floods projects, maintainers clap back—label your demos
TLDR: Discourse’s maintainers warn that AI can churn code fast, but reviewing it is slow, so they want prototypes kept separate from real, ready-to-merge work. Comments split: founders cheer speed and smaller teams, others demand clear rules, and many say the hype is cooling—label your demos or face pushback.
Open-source maintainers at Discourse just dropped a spicy note: AI code-writing tools like Copilot and Claude can spit out giant pull requests (requests to add code) in minutes—but humans still spend hours untangling them. Their fix? A two‑lane road: fun prototype demos you share as links or videos, and real, ready‑for‑review code that follows rules. They even call AI output “alien intelligence,” which a lot of engineers felt in their bones. The mood: stop lobbing “vibe-coded” changes over the fence and start labeling your experiments.
Comments went full popcorn. A startup founder bragged, “I can work with my engineers a lot faster,” and said AI lets them do more with fewer people—cue job-security side‑eye. Others wanted simple rules: “Shouldn’t there be guidelines…?” Meanwhile, one veteran sighed we’re finally “passing the frothing‑at‑the‑mouth stage” of hype. Some memed prototypes as “movie set code”—looks real, collapses if you lean on it. One thread veered into identity-laced snark and was swiftly smacked down. Overall vibe: ship demos, but don’t dump cleanup on maintainers.
Bonus joke: dubbed “dv new my‑experiment” “DIY TikTok for code,” while maintainers begged for “DO NOT MERGE” labels. The community agrees: prototypes are fun, but grown‑up code needs tests, standards, time.
Key Points
- •AI coding tools make code generation cheap but leave code review costly, straining open-source maintainers.
- •Discourse reports increasing volumes of AI-generated contributions and expects similar challenges across open-source projects.
- •Authors propose a binary framework: prototypes versus ready-for-review PRs that meet contribution guidelines.
- •Prototypes are live demos not meeting standards, often lacking tests and potentially introducing security issues and technical debt.
- •Guidelines: don’t open PRs for prototypes; share branches/videos/snippets and disclose AI usage; tools like dv and various AI agents support prototyping.