When the creator of the world's most advanced coding agent speaks, Silicon Valley listens. Boris Cherny, the mastermind behind Anthropic's Claude Code, recently shared his personal terminal setup on X, sparking a viral debate about the future of software development.
Cherny's workflow is deceptively simple yet powerful: he runs five Claude agents in parallel, treating coding like a real-time strategy game. "I run 5 Claudes in parallel in my terminal," he wrote. "I number my tabs 1-5, and use system notifications to know when a Claude needs input." This setup allows one developer to operate with the output of a small engineering team.
Cherny also revealed he exclusively uses Opus 4.5, Anthropic's slowest and smartest model, arguing that the upfront "compute tax" pays off by reducing the time spent correcting errors. "It's the best coding model I've ever used, and even though it's bigger & slower than Sonnet, since you have to steer it less and it's better at tool use, it is almost always faster than using a smaller model in the end," he explained.
To combat AI amnesia, Cherny's team maintains a CLAUDE.md file in their repository. Every time Claude makes a mistake, they add it to the file, turning errors into permanent rules. "Every mistake becomes a rule," noted product leader Aakash Gupta, highlighting how the codebase becomes a self-correcting organism.
Cherny also relies on slash commands and subagents to automate repetitive tasks. A command like /commit-push-pr handles version control bureaucracy with a single keystroke, while specialized subagents simplify code and run end-to-end tests.
The key to Claude Code's success, however, is the verification loop. "Claude tests every single change I land to claude.ai/code using the Claude Chrome extension," Cherny wrote. "It opens a browser, tests the UI, and iterates until the code works and the UX feels good." This ability to verify its own work improves quality by 2-3x, according to Cherny.
The developer community's reaction has been electric. "If you're not reading the Claude Code best practices straight from its creator, you're behind as a programmer," wrote Jeff Tang. Others see Cherny's approach as a turning point, moving AI coding from autocomplete to full-scale labor orchestration. As one user noted, the experience "feels more like Starcraft" than traditional coding.
Cherny's workflow signals a shift: the tools to multiply human output are already here, requiring only a willingness to think differently about how software gets built.