Claude Code Tips From the Guy Who Built It (www.anup.io)

Boris Cherny, a key developer behind Claude Code at Anthropic, has shared insightful tips on optimizing the use of this AI coding tool through Twitter threads. He emphasizes a streamlined approach, suggesting users run multiple instances of Claude Code in parallel—both locally and in the cloud—to enhance productivity. The use of git worktrees allows teams to manage multiple branches more efficiently, while maintaining a shared document, CLAUDE.md, fosters collective learning from mistakes made by the AI, creating a living repository of best practices. Significantly, Boris advocates for a methodology where users stick with a single AI model (Opus 4.5) for a week to assess its effectiveness, and always initiate complex tasks in “Plan Mode.” By employing automation strategies through slash commands and subagents, users can significantly minimize repetitive tasks and leverage Claude Code as a powerful parallel execution engine rather than merely a chatbot. The overarching message is clear: effective configuration, clear communication, and real-time verification are key to maximizing the tool's potential in AI-assisted coding workflows.