June 24, 2026

Ctrl+Alt+Delete the bot flood

PR spam today looks like email spam in the early 2000s

Open-source helpers are drowning in robot-made junk, and commenters are absolutely fed up

TLDR: A fast-growing software project got buried under thousands of low-quality, likely AI-made code submissions, with acceptance rates collapsing. Commenters are split between building trust filters, locking projects down, or banning AI submissions outright — because nobody wants open-source to become the new spam folder.

The big shocker here isn’t just that one popular project went from 2 code submissions a week to 3,400 almost overnight — it’s that the comments section instantly turned into a full-blown “burn it all down” debate. The article argues that coding submissions are starting to look like the email spam plague of the early internet: cheap to send, easy to flood, and suddenly impossible to trust. One person allegedly fired off 106 submissions in a single day, with just three seconds between them, which sounds less like helpful volunteering and more like a robot leaning on the keyboard.

And the crowd? Not subtle. One camp wants reputation systems and gatekeeping, basically a trust score for who gets through the door. Another camp is already beyond compromise: ban all AI-made submissions, period. Commenter runarberg came in swinging, saying AI reviewing AI-created junk is absolutely not the solution. Meanwhile, j2kun offered the most delightfully awkward anti-bot fix imaginable: first-time contributors have to meet a real maintainer in a non-text format before getting accepted — yes, apparently you may need to prove you are a human with vibes.

Then there were the funniest reactions: confusion over who is even making money from this, side-eye at GitHub for not locking things down harder, and a practical suggestion to skip the robot circus entirely and just let people donate computing credits directly to project maintainers. The mood is clear: if the future of open-source help is armies of identical bots submitting the same fixes, the community is deeply unimpressed and already plotting the barricades.

Key Points

  • The article says OpenClaw’s pull request volume rose from 2 per week in December to 3,400 per week by February, while merge rates fell from about 48% to under 9.3%.
  • Many of the new pull requests were described as low-effort and often AI-generated, including one contributor who submitted 106 PRs in one day with a median gap of three seconds.
  • The article argues that PR spam resembles early email spam and suggests similar defenses such as blocklists, filters, and reputation-based systems.
  • OpenClaw data in the article shows higher merge rates for contributors with more prior PR history: 8.2% for first-timers, 10.3% for those with 2–5 PRs, and 18.6% for those with 5+.
  • The article reports that feature PRs on OpenClaw had a 9% merge rate while refactors had a 35% merge rate, and it cites several examples of duplicate fixes or features submitted by different contributors.

Hottest takes

"AI agents who review the slop created by other AI agents is not the answer here" — runarberg
"all new contributors meet a maintainer in a non-textual format" — j2kun
"Where is the profit?" — fecal_henge
Made with <3 by @siedrix and @shesho from CDMX. Powered by Forge&Hive.