{"slug": "boris-chernys-claude-md", "title": "Boris Cherny’s CLAUDE.md", "summary": "This article outlines a structured workflow methodology for AI coding agents, emphasizing the use of \"plan mode\" for complex tasks, strict verification before task completion, and continuous learning through a lessons log. It instructs the agent to demand elegant solutions, fix bugs independently without user hand-holding, and maintain a clean context window by offloading work to subagents. The core principles include planning first, tracking progress, documenting results, and ruthlessly iterating on mistakes to achieve senior developer standards.", "body_md": "CLAUDE.md\n- Enter plan mode for ANY non-trivial task (3+ steps or architectural decisions)\n- If something goes sideways, STOP and re-plan immediately - don't keep pushing\n- Use plan mode for verification steps, not just building\n- Write detailed specs upfront to reduce ambiguity\n- Use subagents liberally to keep main context window clean\n- Offload research, exploration, and parallel analysis to subagents\n- For complex problems, throw more compute at it via subagents\n- One tack per subagent for focused execution\n- After ANY correction from the user: update\ntasks/lessons.md\nwith the pattern\n- Write rules for yourself that prevent the same mistake\n- Ruthlessly iterate on these lessons until mistake rate drops\n- Review lessons at session start for relevant project\n4. Verification Before Done\n- Never mark a task complete without proving it works\n- Diff behavior between main and your changes when relevant\n- Ask yourself: \"Would a staff engineer approve this?\"\n- Run tests, check logs, demonstrate correctness\n5. Demand Elegance (Balanced)\n- For non-trivial changes: pause and ask \"is there a more elegant way?\"\n- If a fix feels hacky: \"Knowing everything I know now, implement the elegant solution\"\n- Skip this for simple, obvious fixes - don't over-engineer\n- Challenge your own work before presenting it\n- When given a bug report: just fix it. Don't ask for hand-holding\n- Point at logs, errors, failing tests - then resolve them\n- Zero context switching required from the user\n- Go fix failing CI tests without being told how\n- Plan First: Write plan to\ntasks/todo.md\nwith checkable items\n- Verify Plan: Check in before starting implementation\n- Track Progress: Mark items complete as you go\n- Explain Changes: High-level summary at each step\n- Document Results: Add review section to\ntasks/todo.md\n- Capture Lessons: Update\ntasks/lessons.md\nafter corrections\n- Simplicity First: Make every change as simple as possible. Impact minimal code.\n- No Laziness: Find root causes. No temporary fixes. Senior developer standards.\n- Minimat Impact: Changes should only touch what's necessary. Avoid introducing bugs.", "url": "https://wpnews.pro/news/boris-chernys-claude-md", "canonical_source": "https://gist.github.com/hqman/e29cb6386c539d795767e8c3fd2c959b", "published_at": "2026-02-23 13:03:34+00:00", "updated_at": "2026-05-23 02:03:26.195674+00:00", "lang": "en", "topics": ["developer-tools", "artificial-intelligence", "large-language-models"], "entities": ["Boris Cherny"], "alternates": {"html": "https://wpnews.pro/news/boris-chernys-claude-md", "markdown": "https://wpnews.pro/news/boris-chernys-claude-md.md", "text": "https://wpnews.pro/news/boris-chernys-claude-md.txt", "jsonld": "https://wpnews.pro/news/boris-chernys-claude-md.jsonld"}}