Skip to content
Anthropic · Claude.dev· Addy Osmani·· 10 h agoPicksAI score60

How the Claude Code team approaches verification for agent-written code

How the Claude Code team thinks about verification

AI summary

The Claude Code team says about 90% of its code is written and reviewed by Claude, and it relies on verification loops to delegate at that scale. In internal testing on structured file generation, a separate grader checking Claude's work against a rubric improved task success by up to 10 points over a standard prompting loop, with the largest gains on the hardest problems. The team favors end-to-end tests, property tests and differential tests, runs a verify skill before commits, sends a check to a human after three failed fix rounds, and keeps checks on a branch the change cannot modify.

Why it matters

The Claude Code team's account of how verification loops let roughly 90% of its code be written and reviewed by Claude offers a transferable method for running many agents in parallel.

Source:Anthropic · Claude.dev · claude.dev