Google open-sources its Elephant-Goldfish AI code review plugin
Google just open-sourced an internal plugin its engineers use to catch AI coding mistakes before they become real bugs. It’s called the Elephant-Goldfish Model plugin, and the approach is worth understanding if you’re running any AI coding workflow.
How It Works
The plugin forces an AI coding agent to write a design document and defend it before it’s allowed to touch any code. That document then gets handed to a completely fresh AI session, one that has never seen the original design conversation, with a single job: find what’s wrong with it.
The name reflects the two-model dynamic. One model holds the full context (elephant). The other comes in cold with no memory of how the design was reasoned through (goldfish). The cold reviewer catches assumptions the original agent couldn’t see because it was too close to its own reasoning.
Why This Matters for Operators
If you’re using AI agents to write production code, the failure mode isn’t usually bad syntax. It’s flawed design decisions that look coherent from inside a single context window. A fresh reviewer with no prior context is better positioned to catch those than the same model that wrote the plan.
This is the same principle behind human code review, applied to AI-to-AI critique. Google is now giving that pattern away for free.
I’m founder of BizStack at Cagri Sarigoz LLC and a passionate advocate for entrepreneurs.
With over 14 years in tech, marketing, and AI, including my role as Head of SEO at CitizenShipper and co-founder of TaleBot at Intale AI, I’m dedicated to sharing genuine, useful product insights and tips.
At BizStack, I aim to cut through the digital noise to provide clear, actionable advice.
And more than all else, I’m a father to a (always) little girl and a husband.
Contact me at [email protected] for assistance.
