TLDRocket
Sign in

Quoting Geoffrey Huntley

Simon Willison’s Weblog Simon Willison

Opinion — commentary, not a factual news event.

Geoffrey Huntley says a pre-commit hook can stop agents from skipping his diagnostics library. His trick turns code review into a hard gate, not a polite suggestion.

Based on reporting by Simon Willison’s Weblog, Simon Willison — read the original for the full story.

Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error

Simon Willison posted a quotation from Geoffrey Huntley on 13 October 2026, and the idea is blunt in the best way. If an agent keeps reaching around a diagnostics library, Huntley says you can try to update AGENTS.md. But there’s still drift. The instructions and the code will slowly stop matching, and the gap will grow.

So he takes a more aggressive route: a pre-commit hook that inspects the git diff for direct calls to observability modules that bypass his diagnostics module. If it finds them, the commit fails. The hook also injects a prompt telling the agent to use the diagnostics library instead. No negotiation. No “please remember this next time.”

That turns a soft policy into something the tooling can actually enforce. Huntley’s punchline is that it makes the setup “ten out of ten.” The quote ends with a familiar complaint dressed up as a joke: nobody should type code by hand anymore. Here’s what’s left.

Willison tags the post with git, ai, generative-ai, llms, coding-agents, Geoffrey Huntley, and agentic-engineering, which is a pretty good map of the moment. The work is no longer just writing code. It’s writing the rails that keep the code, and the agents making it, from wandering off.

My take — AI-written commentary, not fact-checked reporting

This is the real shape of agentic coding: not magic, just enforcement. If the model won’t reliably follow the house rules, make the house rules executable and let git be the bouncer. That’s far less glamorous than “AI engineers itself,” which is exactly why it works.

Read more about this at: Simon Willison’s Weblog

Related stories

The daily briefing

Every AI story that matters, in your inbox by 8am.

TLDRocket reads all relevant sources, removes duplicate coverage, and summarises the day in two minutes. Follow companies and topics for alerts, or get the briefing in Slack. Free, no spam, unsubscribe anytime.