Urgent.News

What's breaking now, across thousands of outlets.

Tech

Zero-hallucination is about schemas, not prompts

Zero-hallucination is about schemas, not prompts The standard advice for getting better output from an AI coding agent is to write a better prompt. Longer prompt, clearer prompt, more examples in the prompt. It helps, up to a point — and then it stops helping, in a very specific way. A prompt can tell an agent what to build. It cannot, by itself, decide the things a codebase actually runs on: the…

Even with a comprehensive specification provided to an AI coding agent, it still has the potential to invent additional elements. The specification serves to determine the core design decisions, such as the tables, foreign keys, data structures, authorization rules, and error states. However, the agent may still need to generate supplementary files like API contracts, Row Level Security policies, and phased build instructions.

These additional components are necessary to ensure the implementation aligns precisely with the specified design, despite the agent's reliance on the provided specification for its primary tasks.

Written by urgent.news from Dev.to's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.

Read the original at dev.to →

More in Tech

639 reads and "no evidence of reading"

On 12 July 2026 a run of six tasks reported 639 file reads. In the same decision, my authorizer rejected the run's answer because there was "no evidence of reading the design document" Both statements…

Bots guess card numbers at online checkouts

Stolen card data is only useful if it still works. So criminals run programs that try card numbers, expiration dates and security codes in large batches, often through ordinary online checkouts.

More from Tuesday 17 November →