My CLAUDE.md workflow orchestration

CLAUDE.md

Workflow Orchestration

1. Plan Mode Default

• Enter plan mode for ANY non-trivial task (3+ steps or architectural decisions)
• If something goes sideways, STOP and re-plan immediately - don't keep pushing
• Use plan mode for verification steps, not just building
• Write detailed specs upfront to reduce ambiguity

2. Subagent Strategy

• Use subagents liberally to keep main context window clean
• Offload research, exploration, and parallel analysis to subagents
• For complex problems, throw more compute at it via subagents
• One task per subagent for focused execution
• Automatically trigger subagent use for: tasks requiring reading multiple large files, extensive research (3+ tool calls),
• or parallel exploration of different approaches

3. Self-Improvement Loop

• After ANY correction from the user: update tasks/lessons.md with the pattern
• Write rules for yourself that prevent the same mistake
• Ruthlessly iterate on these lessons until mistake rate drops
• Review lessons at session start for relevant project
• Note: Create tasks/lessons.md if it doesn't exist at session start

4. Verification Before Done

• Never mark a task complete without proving it works
• Diff behavior between main and your changes when relevant
• Ask yourself: "Would a staff engineer approve this?"
• Run tests, check logs, demonstrate correctness
• Update documentation if changes affect public APIs, user-facing behavior, or setup/configuration

5. Demand Elegance (Balanced)

• For non-trivial changes: pause and ask "is there a more elegant way?"
• If a fix feels hacky: "Knowing everything I know now, implement the elegant solution"
• Skip this for simple, obvious fixes - don't over-engineer
• Challenge your own work before presenting it
• Watch for anti-patterns:
  • Repeating similar code in three or more places → create an abstraction
  • Functions with more than four parameters → reconsider the design
  • Deep nesting (3+ levels) → refactor for clarity
  • Magic numbers or strings → use named constants

6. Autonomous Bug Fixing

• When given a bug report: just fix it. Don't ask for hand-holding
• Point at logs, errors, failing tests - then resolve them
• Zero context switching required from the user
• Go fix failing CI tests without being told how

7. Failure Recovery

• If tests start failing or implementation goes wrong: STOP immediately
• Compare current state with last working version
• Enumerate 2-3 possible solutions with pros/cons
• Choose the simplest path forward
• Avoid entering a fixing loop - verify understanding before continuing
• If fundamentally wrong: acknowledge it, roll back, and re-plan

Task Management

1. Plan First: Write plan to tasks/todo.md with checkable items (create file if doesn't exist)
2. Verify Plans: Check in before starting implementation
3. Track Progress: Mark items complete as you go
4. Explain Changes: High-level summary at each step
5. Document Results: Add review section to tasks/todo.md
6. Capture Lessons: Update tasks/lessons.md after corrections

Technology Context

Stack Preferences:

• Language: TypeScript
• Runtime: Node.js 22+
• Package Manager: pnpm
• Code Quality: ESLint, Prettier
• Testing: Vitest for unit tests

Core Principles

• Simplicity First: Make every change as simple as possible. Impact minimal code.
• No Laziness: Find root causes. No temporary fixes. Senior developer standards.
• Minimal Impact: Changes should only touch what's necessary. Avoid introducing bugs.
• Incremental Progress: Favor small, verifiable steps over large risky changes. Each step should be testable and reviewable. Baby steps are not just acceptable but preferred.