workflow-orchestration（工作流程编排）

Workflow Orchestration

Apply these practices for disciplined, high-quality task execution.

Quick Reference

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
• Use plan mode for verification steps, not just building
• Write detailed specs upfront to reduce ambiguity

2. Subagent Strategy

Keep the 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

Template:

Agent({
  description: "Search auth patterns",
  prompt: "Find all authentication middleware in src/ and list file paths with a one-line summary of each."
})

3. Self-Improvement Loop

After ANY correction from the user:

1. Update tasks/lessons.md with the pattern
2. Write rules that prevent the same mistake
3. Review lessons at session start

Inline example (full template in references/lessons-format.md):

## 2024-01-10 - Testing
**Mistake**: Mocks hid real API behavior; production broke
**Pattern**: Over-mocking created false confidence
**Rule**: Include at least one integration test hitting real services
**Applied**: All API endpoints, external service integrations

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 corrections
• If verification fails → return to the plan step and revise before retrying

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

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

Task Management Protocol

1. Plan First: Write plan to tasks/todo.md with checkable items
2. Verify Plan: Check in before starting implementation
3. Track Progress: Mark items complete as you go
4. Explain Changes: High-level summary at each step
5. Verify Results: Run tests and confirm behavior before marking done
6. Handle Failures: If verification fails → re-plan (step 1), don't push through
7. Document Results: Add review to tasks/todo.md
8. Capture Lessons: Update tasks/lessons.md after corrections

Inline example (full templates in references/task-templates.md):

# Task: Fix memory leak in WebSocket handler
## Plan
- [ ] Reproduce the issue locally
- [ ] Identify leak source with profiler
- [ ] Implement fix
- [ ] Verification: Memory stable over 1000 connections

Core Principles

• Simplicity First: Make every change as simple as possible
• No Laziness: Find root causes. No temporary fixes. Senior developer standards
• Minimal Impact: Only touch what's necessary. Avoid introducing bugs