ralph-prompt-multi-task（拉尔夫提示多任务）

Ralph Prompt Generator: Multi-Task

Overview

Generates structured prompts for multiple related tasks that need to be completed in sequence or parallel. Uses phase-based organization with milestone tracking, cumulative success criteria, and clear progression through tasks.

Best For:

• CRUD operation implementations (Create, Read, Update, Delete)
• Multi-step feature development
• Database migrations with multiple stages
• API endpoint suites
• Test suite creation for multiple modules
• Multi-file refactoring projects

Ralph Philosophy: This generator embraces the principle that failures are deterministic and fixable. Each phase iteration learns from previous attempts. Don't fear failures—they're expected and provide data for improvement through prompt tuning.

Quick Start

Input Required:

1. List of tasks to complete
2. Task dependencies (which tasks depend on others)
3. Success criteria for each task
4. Final completion promise

Generate prompt with:

Generate a Ralph multi-task prompt for:
Tasks:
1. [Task 1]
2. [Task 2]
3. [Task 3]
Dependencies: [Task 2 depends on Task 1, etc.]
Final promise: [COMPLETION_PHRASE]

Prompt Generation Workflow

Step 1: Map the Tasks

Create a task inventory:

Step 2: Define Phases

Group tasks into logical phases:

Phase Structure:

• Phase 1: Foundation - Independent setup tasks, no dependencies
• Phase 2: Core Implementation - Main functionality, depends on Phase 1
• Phase 3: Enhancement - Additional features, depends on Phase 2
• Phase 4: Validation - Testing, cleanup, verification

Step 3: Create Phase Milestones

Each phase needs:

• Clear entry criteria (what must be done to start)
• Specific tasks within the phase
• Exit criteria (what proves phase is complete)
• Phase milestone marker

Step 4: Structure the Prompt

Use this template:

# Multi-Task: [Overall Objective]

## Overview
[1-2 sentences describing the overall goal and scope]

## Task Inventory
| # | Task | Phase | Dependencies | Verification |
|---|------|-------|--------------|--------------|
| 1 | [Task 1] | 1 | None | [Verification] |
| 2 | [Task 2] | 2 | Task 1 | [Verification] |
| 3 | [Task 3] | 2 | Task 1 | [Verification] |
| 4 | [Task 4] | 3 | Tasks 2, 3 | [Verification] |

## Phase 1: [Phase Name] (Foundation)

### Objective
[What this phase accomplishes]

### Tasks
1. **[Task 1 Name]**
   - Requirements: [Specific requirements]
   - Files: [Files to create/modify]
   - Verification: [How to verify]

### Phase 1 Success Criteria
- [ ] [Criterion 1]
- [ ] [Criterion 2]
- [ ] Tests pass: `[test command]`

### Phase 1 Checkpoint
When Phase 1 criteria met, document:

PHASE 1 COMPLETE:

• [Task 1]: Done - [evidence]
• Tests: [pass/fail status]

Continue to Phase 2.

---

## Phase 2: [Phase Name] (Core)

### Objective
[What this phase accomplishes]

### Prerequisites
- Phase 1 complete
- [Any specific requirements]

### Tasks
2. **[Task 2 Name]**
   - Requirements: [Specific requirements]
   - Files: [Files to create/modify]
   - Verification: [How to verify]

3. **[Task 3 Name]**
   - Requirements: [Specific requirements]
   - Files: [Files to create/modify]
   - Verification: [How to verify]

### Phase 2 Success Criteria
- [ ] [Criterion 1]
- [ ] [Criterion 2]
- [ ] [Criterion 3]
- [ ] Tests pass: `[test command]`

### Phase 2 Checkpoint
When Phase 2 criteria met, document:

PHASE 2 COMPLETE:

• [Task 2]: Done - [evidence]
• [Task 3]: Done - [evidence]
• Tests: [pass/fail status]

Continue to Phase 3.

---

## Phase 3: [Phase Name] (Enhancement/Validation)

### Objective
[What this phase accomplishes]

### Prerequisites
- Phases 1-2 complete
- [Any specific requirements]

### Tasks
4. **[Task 4 Name]**
   - Requirements: [Specific requirements]
   - Files: [Files to create/modify]
   - Verification: [How to verify]

### Phase 3 Success Criteria
- [ ] [Criterion 1]
- [ ] [Criterion 2]
- [ ] All tests pass: `[test command]`

---

## Final Verification

Run complete verification:

All tests

[full test command]

Integration check

[integration verification]

Lint/Type check

[lint command]

## Completion Checklist

- Phase 1: All tasks complete and verified
- Phase 2: All tasks complete and verified
- Phase 3: All tasks complete and verified
- All tests passing
- No regressions
- [Any additional requirements]

## Completion

When ALL phases complete and all checklist items verified:

Output: [COMPLETION_PHRASE]

## Progress Tracking

As you work, track progress in this format:

PROGRESS: Phase 1: [COMPLETE/IN PROGRESS] - [X/Y tasks] Phase 2: [COMPLETE/IN PROGRESS/NOT STARTED] - [X/Y tasks] Phase 3: [COMPLETE/IN PROGRESS/NOT STARTED] - [X/Y tasks] Overall: [X/Y tasks complete]

## If Stuck

- Document which phase/task is blocking
- List what's been tried
- Identify the specific blocker
- Do NOT skip tasks or output false promise

Example Prompts

Example 1: CRUD API Implementation

Input:

• Tasks: Create, Read (single), Read (list), Update, Delete for Products
• Dependencies: Create first, then others can be parallel
• Promise: CRUD_API_COMPLETE

Generated Prompt:

# Multi-Task: Product CRUD API Implementation

## Overview
Implement complete CRUD operations for Products resource with RESTful endpoints, validation, and comprehensive tests.

## Task Inventory
| # | Task | Phase | Dependencies | Verification |
|---|------|-------|--------------|--------------|
| 1 | Product model/schema | 1 | None | Model exists, validates |
| 2 | POST /products (Create) | 2 | Task 1 | Returns 201, creates record |
| 3 | GET /products/:id (Read) | 2 | Task 1 | Returns 200, correct data |
| 4 | GET /products (List) | 2 | Task 1 | Returns array, pagination |
| 5 | PUT /products/:id (Update) | 3 | Tasks 2,3 | Returns 200, updates record |
| 6 | DELETE /products/:id | 3 | Tasks 2,3 | Returns 204, removes record |
| 7 | Integration tests | 4 | Tasks 2-6 | All scenarios pass |

## Phase 1: Foundation

### Objective
Create Product model with validation and database schema.

### Tasks
1. **Product Model**
   - Requirements:
     - Fields: id, name, description, price, category, createdAt, updatedAt
     - Validation: name required, price > 0, category from enum
   - Files: `src/models/product.ts`, `src/schemas/product.ts`
   - Verification: Model compiles, validation tests pass

### Phase 1 Success Criteria
- [ ] Product model created with all fields
- [ ] Validation rules implemented
- [ ] Database migration/schema created
- [ ] Model tests pass: `npm test -- --grep "Product model"`

### Phase 1 Checkpoint

PHASE 1 COMPLETE:

• Product model: Done - all fields, validation working
• Tests: All model tests passing

---

## Phase 2: Core CRUD Operations

### Objective
Implement Create, Read (single), and Read (list) endpoints.

### Prerequisites
- Phase 1 complete (Product model exists)

### Tasks
2. **POST /products (Create)**
   - Requirements:
     - Accept JSON body with product data
     - Validate input using schema
     - Return 201 with created product
     - Return 400 for validation errors
   - Files: `src/routes/products.ts`, `src/controllers/products.ts`
   - Verification: `curl -X POST -H "Content-Type: application/json" -d '{"name":"Test","price":9.99,"category":"electronics"}' localhost:3000/products`

3. **GET /products/:id (Read Single)**
   - Requirements:
     - Return 200 with product data
     - Return 404 if not found
   - Files: `src/routes/products.ts`, `src/controllers/products.ts`
   - Verification: `curl localhost:3000/products/1`

4. **GET /products (List)**
   - Requirements:
     - Return array of products
     - Support pagination: page, limit query params
     - Return metadata: total, page, limit, totalPages
   - Files: `src/routes/products.ts`, `src/controllers/products.ts`
   - Verification: `curl "localhost:3000/products?page=1&limit=10"`

### Phase 2 Success Criteria
- [ ] POST /products creates product, returns 201
- [ ] POST /products returns 400 for invalid data
- [ ] GET /products/:id returns product
- [ ] GET /products/:id returns 404 for missing
- [ ] GET /products returns paginated list
- [ ] All endpoint tests pass: `npm test -- --grep "products"`

### Phase 2 Checkpoint

PHASE 2 COMPLETE:

• POST /products: Done - creates products, validates input
• GET /products/:id: Done - returns product or 404
• GET /products: Done - paginated list working
• Tests: All passing

---

## Phase 3: Update and Delete Operations

### Objective
Complete CRUD with Update and Delete endpoints.

### Prerequisites
- Phase 2 complete (Create and Read working)

### Tasks
5. **PUT /products/:id (Update)**
   - Requirements:
     - Accept JSON body with update data
     - Partial updates allowed
     - Return 200 with updated product
     - Return 404 if not found
     - Return 400 for validation errors
   - Verification: `curl -X PUT -H "Content-Type: application/json" -d '{"price":19.99}' localhost:3000/products/1`

6. **DELETE /products/:id**
   - Requirements:
     - Return 204 on success (no content)
     - Return 404 if not found
     - Actually remove record from database
   - Verification: `curl -X DELETE localhost:3000/products/1`

### Phase 3 Success Criteria
- [ ] PUT /products/:id updates product
- [ ] PUT returns 404 for missing, 400 for invalid
- [ ] DELETE /products/:id removes product
- [ ] DELETE returns 404 for missing
- [ ] All endpoint tests pass

### Phase 3 Checkpoint

PHASE 3 COMPLETE:

• PUT /products/:id: Done - updates work
• DELETE /products/:id: Done - deletes work
• Tests: All passing

---

## Phase 4: Integration & Validation

### Objective
Complete test coverage and integration verification.

### Tasks
7. **Integration Tests**
   - Requirements:
     - Full CRUD flow test (create, read, update, delete)
     - Error handling tests (400, 404 scenarios)
     - Pagination tests
     - Edge cases (empty database, special characters)
   - Files: `tests/integration/products.test.ts`

### Phase 4 Success Criteria
- [ ] Integration tests cover full CRUD flow
- [ ] Error scenarios tested
- [ ] All tests passing: `npm test`
- [ ] No TypeScript errors: `npm run typecheck`

---

## Final Verification

All tests

npm test

Type check

npm run typecheck

Manual CRUD flow test

curl -X POST -H "Content-Type: application/json" -d '{"name":"Test Product","price":29.99,"category":"electronics"}' localhost:3000/products curl localhost:3000/products curl localhost:3000/products/1 curl -X PUT -H "Content-Type: application/json" -d '{"price":39.99}' localhost:3000/products/1 curl -X DELETE localhost:3000/products/1

## Completion Checklist

- Phase 1: Product model complete
- Phase 2: Create, Read single, Read list working
- Phase 3: Update, Delete working
- Phase 4: Integration tests complete
- All tests passing
- Manual verification successful

## Completion

When ALL phases complete and all checklist items verified:

Output: CRUD_API_COMPLETE

## Progress Tracking

PROGRESS: Phase 1: [status] - 1/1 tasks Phase 2: [status] - 3/3 tasks Phase 3: [status] - 2/2 tasks Phase 4: [status] - 1/1 tasks Overall: X/7 tasks complete

## If Stuck

- Document which phase/task is blocking
- Note specific error messages
- List approaches tried
- Do NOT skip tasks or output false promise

Example 2: CI/CD Pipeline Setup

Input:

• Tasks: Lint, Test, Build, Deploy staging, Deploy prod
• Dependencies: Sequential progression
• Promise: PIPELINE_COMPLETE

Generated Prompt:

# Multi-Task: CI/CD Pipeline Implementation

## Overview
Set up complete CI/CD pipeline with linting, testing, building, and staged deployments using GitHub Actions.

## Task Inventory
| # | Task | Phase | Dependencies | Verification |
|---|------|-------|--------------|--------------|
| 1 | Workflow file structure | 1 | None | File exists |
| 2 | Lint job | 2 | Task 1 | Workflow runs lint |
| 3 | Test job | 2 | Task 2 | Tests run in CI |
| 4 | Build job | 3 | Task 3 | Build artifacts created |
| 5 | Deploy staging | 4 | Task 4 | Staging deployment works |
| 6 | Deploy production | 4 | Task 5 | Prod deployment works |

## Phase 1: Workflow Foundation

### Tasks
1. **Workflow File Structure**
   - Create `.github/workflows/ci.yml`
   - Define triggers (push to main, PRs)
   - Set up job structure

### Phase 1 Success Criteria
- [ ] Workflow file exists and is valid YAML
- [ ] Triggers defined for push and PR

---

## Phase 2: Quality Gates

### Tasks
2. **Lint Job**
   - Run linter on all source files
   - Fail pipeline if lint errors

3. **Test Job**
   - Depends on lint passing
   - Run full test suite
   - Upload coverage report

### Phase 2 Success Criteria
- [ ] Lint job runs and catches errors
- [ ] Test job runs full suite
- [ ] Pipeline fails if tests fail

---

## Phase 3: Build

### Tasks
4. **Build Job**
   - Depends on tests passing
   - Create production build
   - Upload build artifacts

### Phase 3 Success Criteria
- [ ] Build creates production artifacts
- [ ] Artifacts uploaded and accessible

---

## Phase 4: Deployments

### Tasks
5. **Deploy Staging**
   - Automatic on main branch
   - Deploy to staging environment

6. **Deploy Production**
   - Manual approval required
   - Deploy to production

### Phase 4 Success Criteria
- [ ] Staging auto-deploys on main
- [ ] Production requires manual trigger
- [ ] Both environments update correctly

---

## Completion
When all phases verified:

Output: <promise>PIPELINE_COMPLETE</promise>

Best Practices

Task Organization

• Group related tasks into phases
• Clear dependencies between phases
• Independent tasks within phases can be parallel

Phase Design

• Each phase has clear entry/exit criteria
• Document checkpoint at each phase boundary
• Phase names should reflect purpose (Foundation, Core, Enhancement, Validation)

Progress Tracking

• Update progress marker after each task
• Document evidence at checkpoints
• Don't skip phases even if tasks seem simple

DO:

• Create clear phase boundaries
• Define dependencies explicitly
• Include verification for each task
• Track progress systematically
• Checkpoint after each phase

DON'T:

• Skip phases or tasks
• Leave dependencies unclear
• Forget phase checkpoints
• Output promise before all phases complete

Integration with Ralph Loop

/ralph-wiggum:ralph-loop "[paste generated prompt]" --completion-promise "YOUR_PROMISE" --max-iterations 50

Recommended iterations by complexity:

• 4-5 tasks: --max-iterations 35-45
• 6-8 tasks: --max-iterations 50-70
• 9+ tasks: --max-iterations 80-100

For single-task prompts, see ralph-prompt-single-task. For project-level prompts, see ralph-prompt-project. For research/analysis prompts, see ralph-prompt-research.