Coding Models Are Doing Too Much | Hacker News
TL;DR AI
2 min readKey summary
Hacker News users debated whether coding agents speed up development or encourage dangerous over-automation.
The concern is that agents can edit multiple files, run tests, and deploy code with too little human oversight.
Commenters called for manual approval, code review, and keeping developers understanding the system.
Security risks included accidental database wipes and exposure of AWS credentials, .env files, or deployment targets.
The core tradeoff is productivity gains versus lost control, hidden behavior, and safety blind spots.



