fortunto2/rust-code/plan

WORKFLOW:

1. READ CLAUDE.md + README.md — architecture, stack, constraints
2. TREE root level=3 — understand project structure
3. SEARCH for keywords from task description — find affected files
4. READ existing plans — list("docs/plan") to check overlap
5. GENERATE track ID: {shortname}_{YYYYMMDD} (kebab-case)
6. WRITE docs/plan/{trackId}/spec.md:

`` # Specification: {Title} **Track ID:** {trackId} **Type:** Feature|Bug|Refactor|Chore **Created:** {date} ## Summary — 1-2 paragraphs from research ## Acceptance Criteria — 3-8 concrete checkboxes ## Dependencies ## Out of Scope ## Technical Notes — architecture decisions, reusable code found ``

1. WRITE docs/plan/{trackId}/plan.md:

``` # Implementation Plan: {Title} Spec: spec.md ## Phase 1: {Name} ### Tasks

• [ ] Task 1.1: {description with file paths}

### Verification

• [ ] {check}

## Phase N: Docs & Cleanup

• [ ] Update CLAUDE.md
• [ ] Remove dead code

## Context Handoff ### Key Files — from research ### Decisions Made — why X over Y ### Risks ```

1. UPDATE_PLAN with phases as steps (pending/in_progress/completed)
2. FINISH with: "Track {trackId}: {N} phases, {N} tasks"

RULES:

• Every task mentions specific FILE PATHS (from research, not guessed)
• 5-15 tasks total across 2-4 phases
• Last phase always "Docs & Cleanup"
• Every acceptance criterion maps to at least one task
• Tasks are atomic — one commit each

WRONG: Plan without reading code first (guessing file paths) CORRECT: research → spec → plan → update_plan → finish

EXAMPLE — Plan a new feature: Instruction: "plan adding WebSocket support"

1. read("CLAUDE.md") → understand architecture
2. tree("/", 3) → find relevant modules
3. search("/", "websocket|ws|socket") → check existing code
4. write("docs/plan/websocket-support_20260412/spec.md", ...)
5. write("docs/plan/websocket-support_20260412/plan.md", ...)
6. update_plan(plan: [{step: "Phase 1: Protocol", status: "pending"}, ...])
7. finish("Track websocket-support_20260412: 3 phases, 8 tasks")