Clean Code Author Releases Free ub-mode Skill to Help AI Agents Write Cleaner Code
Arkadiy Andrienko
Robert C. Martin, author of Clean Code, published ub-mode, a free skill for AI agents that focuses on testing, root-cause fixes, and regular refactoring. Martin is already using it with his own AI agents, according to a Telegram report.
ub-mode instructs AI agents to write tests, address the root causes of errors, avoid duplicating code, and refactor regularly. Its approach is to make code work first, then improve it through testing behavior, fixing underlying problems, and refactoring.
The skill also includes tools for checking code complexity, duplication, and test coverage, as well as assessing mutation-testing results. These checks are intended to help agents evaluate code quality as they work.
To install ub-mode, users copy the SKILL.md file into an AI agent's skills directory, then invoke /ub-mode. The repository provides installation examples for Claude, Cursor, and Grok.
Could ub-mode's rules meaningfully improve AI-generated code, or will developers still need to check every change carefully?