software-quality
1 articles
Can AI Test Itself? — From Claude Code's Zero Tests to Self-Testing Agents
Claude Code has 512K lines of TypeScript, 64K lines of production code, and zero tests. The sharper question is not why Anthropic skipped tests, but why it did not use its own AI coding tool to write them. Can the same brain write and grade the exam?