一键导入
task-breakdown
Use when converting architecture plans into implementation phases. Creates independent, bite-sized phase files that can be executed separately.
用 Codex 或 Claude 帮你安装 复制这段 Prompt,粘贴到 Codex、Claude 或其他助手里,让它检查 Skill 页面并帮你完成安装。
菜单
Use when converting architecture plans into implementation phases. Creates independent, bite-sized phase files that can be executed separately.
用 Codex 或 Claude 帮你安装 复制这段 Prompt,粘贴到 Codex、Claude 或其他助手里,让它检查 Skill 页面并帮你完成安装。
You MUST use this before any creative work - creating features, building components, adding functionality, or modifying behavior. Explores user intent, requirements and design before implementation.
Use when encountering any bug, test failure, or unexpected behavior, before proposing fixes
Extract design assets and metadata from Figma using the Figma REST API. Supports exporting frames/components as images, extracting node metadata, design tokens, and file structure. Use with ai-multimodal skill for comprehensive UI research.
Comprehensive codebase documentation generator following a layered methodology. This skill should be used when scanning and documenting a codebase for the first time, when creating onboarding documentation for new developers, when generating architecture overviews, walkthroughs, and API references. Supports README generation, architecture diagrams, entry point documentation, pattern guides, and edge case documentation.
Generate structured handoff summaries between workflow phases. Use when transitioning between RQPIV phases.
Process and generate multimedia content using Google Gemini API. Capabilities include analyze audio files (transcription with timestamps, summarization, speech understanding, music/sound analysis up to 9.5 hours), understand images (captioning, object detection, OCR, visual Q&A, segmentation), process videos (scene detection, Q&A, temporal analysis, YouTube URLs, up to 6 hours), extract from documents (PDF tables, forms, charts, diagrams, multi-page), generate images (text-to-image, editing, composition, refinement). Use when working with audio/video files, analyzing images or screenshots, processing PDF documents, extracting structured data from media, creating images from text prompts, or implementing multimodal AI features. Supports multiple models (Gemini 2.5/2.0) with context windows up to 2M tokens.
基于 SOC 职业分类
| name | task-breakdown |
| description | Use when converting architecture plans into implementation phases. Creates independent, bite-sized phase files that can be executed separately. |
Transform architecture plans into independent implementation phases. Each phase is a separate file that can be executed without waiting for other phases.
Core principle: Independent phases first, dependent phases last.
plans/sessions/{session}/plans/
├── implementation.md # Master index (lightweight)
├── phases/
│ ├── phase-01-foundation.md # Independent - implement first
│ ├── phase-02-core-models.md # Independent - implement first
│ ├── phase-03-api-routes.md # Independent - implement first
│ ├── phase-04-ui-components.md # Depends on models
│ └── phase-05-integration.md # Depends on all above
INDEPENDENT PHASES → TOP (implement first)
DEPENDENT PHASES → BOTTOM (implement last)
Why? Independent phases can be:
| Type | Description | Position |
|---|---|---|
| Foundation | Config, utils, types | Top (first) |
| Core | Models, services | Top |
| Feature | Specific functionality | Middle |
| Integration | Connecting components | Bottom |
| Polish | Tests, docs, cleanup | Bottom (last) |
The master file is lightweight - just an index:
# Implementation Plan: [Feature]
**Session:** {session-id}
**Status:** Proposed | In Progress | Completed
## Phase Summary
| Phase | Name | Status | Dependencies |
|-------|------|--------|--------------|
| 1 | Foundation | Pending | None |
| 2 | Core Models | Pending | None |
| 3 | API Routes | Pending | None |
| 4 | UI Components | Pending | Phase 2 |
| 5 | Integration | Pending | Phase 1-4 |
## Execution Order
**Can implement in parallel:**
- Phase 1, 2, 3 (no dependencies)
**Must wait:**
- Phase 4 → after Phase 2
- Phase 5 → after all phases
## Phase Files
- [Phase 1: Foundation](phases/phase-01-foundation.md)
- [Phase 2: Core Models](phases/phase-02-core-models.md)
- [Phase 3: API Routes](phases/phase-03-api-routes.md)
- [Phase 4: UI Components](phases/phase-04-ui-components.md)
- [Phase 5: Integration](phases/phase-05-integration.md)
Each phase file is self-contained:
# Phase [N]: [Phase Name]
**Dependencies:** None | Phase [X], [Y]
**Can Start:** Immediately | After Phase [X]
**Estimated Tasks:** [N]
## Objective
[One sentence: what this phase accomplishes independently]
## Entry Criteria
- [ ] [What must be true before starting]
## Tasks
### Task [N.1]: [Task Name]
**Size:** XS | S | M | L
**File:** `path/to/file.ts`
[Task details using task-template.md format]
---
### Task [N.2]: [Task Name]
...
## Exit Criteria
- [ ] All tasks complete
- [ ] Verification passes
- [ ] No impact on other phases
## Verification
```bash
# Phase-specific verification
[commands]
## Dependency Sorting Algorithm
**Step 1:** List all components from architecture
**Step 2:** For each component, identify what it imports/uses
**Step 3:** Score by dependency count:
- 0 dependencies = Phase 1 (top)
- 1-2 dependencies = Phase 2-3 (middle)
- 3+ dependencies = Last phases (bottom)
**Step 4:** Within same dependency count, order by:
1. Config/Types first (foundational)
2. Services/Models second
3. Controllers/UI third
4. Integration/E2E last
## Task Criteria (Unchanged)
Tasks within phases must be:
| Criterion | Description |
|-----------|-------------|
| **Single-concern** | One logical change |
| **Verifiable** | Clear success criteria |
| **Bounded** | Defined start/end |
| **Sized** | XS, S, M, L (no XL) |
**Rule:** XL tasks must be split further.
## File Operations
| Action | Format |
|--------|--------|
| CREATE | `path/to/new/file` + pattern to follow |
| MODIFY | `path/to/file` + line range + before/after |
| DELETE | `path/to/file` + reason + dependency check |
| MOVE | `from` → `to` + import updates |
## Verification Per Task
| Type | Command | Use Case |
|------|---------|----------|
| Type Check | `npm run typecheck` | TypeScript |
| Lint | `npm run lint` | Style |
| Unit Test | `npm test -- [pattern]` | Logic |
| Build | `npm run build` | Compilation |
## Common Mistakes
| Mistake | Fix |
|---------|-----|
| One massive implementation.md | Split into phase files |
| Dependent phases at top | Reorder: independent first |
| Phase depends on multiple others | Move to bottom |
| XL tasks | Break into S/M tasks |
| Vague verification | Specific commands |
## Output Checklist
Before presenting plan:
- [ ] Implementation.md is index-only (< 50 lines)
- [ ] Each phase has separate file
- [ ] Independent phases listed first
- [ ] Dependent phases at bottom
- [ ] Each phase is self-contained
- [ ] Tasks are appropriately sized
- [ ] Verification commands provided