Claude Code team details evidence-based verification for coding agents
The Claude Code team says Claude writes and reviews about 90% of its code, making evidence-based verification essential at that scale. The team describes using end-to-end, property and differential tests, separate graders and human escalation to check agent work. In internal testing, a separate grader improved task success by up to 10 points over a standard prompting loop, according to the team's account.
The workflow offers a concrete example of how coding agents can be delegated more work while retaining automated checks, human review for persistent failures and safeguards against unreliable verification.
How the Claude Code team thinks about verification / claude.dev Blog
Engineering
How the Claude Code team thinks about verification
How end-to-end, property and differential tests let coding agents verify their work with evidence, and what the Claude Code team does to keep its verifiers honest.
AUTHOR
Addy Osmani