create-task-batch
Installation
SKILL.md
/create-task-batch - Interactive Task List Creator (v2.88)
Interactive wizard for creating PRD documents and task lists optimized for /task-batch execution. Uses guided questioning to extract precise requirements, priorities, and dependencies.
Quick Start
# Start interactive wizard
/create-task-batch
# With initial description
/create-task-batch "User authentication feature"
# For specific project area
/create-task-batch "API refactoring project"
Workflow
+------------------------------------------------------------------+
| CREATE-TASK-BATCH WORKFLOW |
+------------------------------------------------------------------+
| |
| +----------+ +-----------+ +---------------+ |
| | GATHER |--->| DECOMPOSE |--->| PRIORITIZE | |
| | Context | | Into Tasks| | & Dependencies| |
| +----------+ +-----------+ +-------+-------+ |
| | |
| +--------------------------------------------v---------------+ |
| | VALIDATE | |
| | - Each task has acceptance criteria | |
| | - Dependencies are resolvable | |
| | - Priorities are assigned | |
| +---------------------------+-------------------------------+ |
| | |
| +---------------------------v-------------------------------+ |
| | OUTPUT | |
| | - PRD file (.prq.md) | |
| | - Task list (tasks.md) | |
| | - JSON config (optional) | |
| +------------------------------------------------------------+ |
| |
+------------------------------------------------------------------+
Questioning Phases
Phase 1: PROJECT CONTEXT
AskUserQuestion:
questions:
- question: "What is the main goal of this batch?"
header: "Goal"
options:
- label: "New Feature"
description: "Implement a complete new feature"
- label: "Bug Fixes"
description: "Fix multiple related bugs"
- label: "Refactoring"
description: "Restructure existing code"
- label: "Documentation"
description: "Create or update documentation"
- question: "What is the project area?"
header: "Area"
options:
- label: "Backend API"
- label: "Frontend UI"
- label: "Database"
- label: "Infrastructure"
- label: "Full Stack"
- question: "Does the project have a DESIGN.md (design system)?"
header: "Design System"
options:
- label: "Yes, use existing DESIGN.md"
description: "Load design tokens for frontend consistency"
- label: "No, create one with /design-system init"
description: "Initialize a 9-section design system first"
- label: "Not applicable (backend only)"
description: "Skip design system for non-frontend work"
- question: "What is the target timeframe?"
header: "Timeframe"
options:
- label: "Single session"
- label: "1-2 days"
- label: "1 week"
- label: "Sprint (2 weeks)"
Phase 2: FEATURE DETAILS
AskUserQuestion:
questions:
- question: "Describe the main feature/functionality in detail"
header: "Description"
multiSelect: false
# User provides free-form description
- question: "What are the MUST-HAVE components?"
header: "Must Have"
multiSelect: true
options:
- label: "Authentication"
- label: "Data validation"
- label: "Error handling"
- label: "Logging"
- label: "Testing"
- label: "Documentation"
- question: "What are NICE-TO-HAVE components?"
header: "Nice to Have"
multiSelect: true
options:
- label: "Performance optimization"
- label: "Caching"
- label: "Rate limiting"
- label: "Monitoring"
- label: "CI/CD integration"
Phase 3: TASK DECOMPOSITION
# Automatic decomposition based on responses
Task Decomposition Rules:
- Backend Feature:
1. Create data models/schemas
2. Implement business logic
3. Create API endpoints
4. Add validation
5. Write tests
6. Update documentation
- Frontend Feature:
1. Create UI components
2. Implement state management
3. Connect to API
4. Add error handling
5. Write tests
6. Update documentation
- Refactoring:
1. Analyze current implementation
2. Design new structure
3. Create new modules
4. Migrate existing code
5. Update tests
6. Remove deprecated code
Phase 4: PRIORITY & DEPENDENCIES
AskUserQuestion:
questions:
- question: "How should tasks be prioritized?"
header: "Priority"
options:
- label: "Auto (by dependencies)"
- label: "Manual (I will specify)"
- label: "By effort (easiest first)"
- label: "By importance (critical first)"
- question: "Should execution stop on first failure?"
header: "Stop on Fail"
options:
- label: "Yes - Stop entire batch"
- label: "No - Continue with remaining"
Phase 5: ACCEPTANCE CRITERIA (MANDATORY FOR EACH TASK)
CRITICAL: Every task MUST have explicit completion validation criteria. No task can be executed without defined acceptance criteria.
# MANDATORY: For EACH generated task, ask completion criteria
FOR EACH task IN task_list:
AskUserQuestion:
questions:
- question: "Define MANDATORY completion criteria for '[TASK_NAME]':"
header: "Criteria"
# MANDATORY - Cannot proceed without answer
- question: "How will we VERIFY this task is complete?"
header: "Verification"
options:
- label: "Tests pass"
description: "Unit/integration tests must pass"
- label: "File exists"
description: "Specific file(s) created/modified"
- label: "Command succeeds"
description: "Specific command runs without error"
- label: "Manual review"
description: "Human reviews and approves"
- question: "What files/modules are affected by '[TASK_NAME]'?"
header: "Files"
# Optional but recommended
# VALIDATION CHECK
IF any_task_missing_criteria:
BLOCK: "Cannot create batch - all tasks MUST have completion criteria"
ASK: Request missing criteria before proceeding
Task Completion Criteria Template
Each task in the output MUST include:
task:
id: "task-001"
description: "Create OAuth2 service module"
priority: 1
completion_criteria: # MANDATORY
- criteria: "File src/auth/oauth2.service.ts exists"
verification: "file_exists"
- criteria: "Google OAuth client configured"
verification: "code_contains"
pattern: "GoogleAuthProvider"
- criteria: "Unit tests pass"
verification: "command"
command: "npm test -- oauth2.service.spec.ts"
- criteria: "No TypeScript errors"
verification: "command"
command: "tsc --noEmit"
dependencies: []
Output Templates
Template 1: PRD Format (.prq.md)
# Feature: {FEATURE_NAME}
**Created**: {TIMESTAMP}
**Version**: 2.88
**Timeframe**: {TIMEFRAME}
## Priority: {PRIORITY}
## Overview
{DESCRIPTION}
## Tasks
{#for task in tasks}
- [ ] {task.priority}: {task.description}
{#endfor}
## Dependencies
{#for dep in dependencies}
- {dep.task} depends on {dep.depends_on}
{#endfor}
## Acceptance Criteria
{#for task in tasks}
### {task.name}
{#for criteria in task.criteria}
- {criteria}
{#endfor}
{#endfor}
## Technical Notes
{TECHNICAL_NOTES}
## Risks
{#for risk in risks}
- {risk}
{#endfor}
Template 2: Task List Format (tasks.md)
# Batch: {BATCH_NAME}
**Created**: {TIMESTAMP}
**Config**: stop_on_failure={STOP_ON_FAILURE}, auto_commit=true
## Tasks
{#for task in tasks}
{task.number}. [{task.priority}] {task.description}
- Files: {task.files}
- Criteria: {task.criteria}
{#endfor}
## Execution Order
{EXECUTION_ORDER}
Template 3: JSON Format
{
"batch_name": "{BATCH_NAME}",
"created": "{TIMESTAMP}",
"config": {
"stop_on_failure": {STOP_ON_FAILURE},
"auto_commit": true,
"teammates": ["coder", "reviewer"]
},
"tasks": [
{
"id": "task-001",
"description": "{description}",
"priority": 1,
"dependencies": [],
"acceptance_criteria": ["{criteria}"],
"files": ["{files}"]
}
]
}
CLI Usage
# Interactive mode (all questions)
ralph create-batch
# With initial description
ralph create-batch "Implement user authentication"
# Quick mode (minimal questions)
ralph create-batch --quick "Add API rate limiting"
# Output format selection
ralph create-batch --format json
ralph create-batch --format prd
ralph create-batch --format tasks
# Save to specific location (default: .claude/plans/ in project root)
ralph create-batch --output .claude/plans/auth.prq.md
Integration with /task-batch
After creating the task list:
# Step 1: Create the batch
/create-task-batch "User authentication"
# Step 2: Review generated PRD
# File saved to: .claude/plans/user-authentication.prq.md
# Step 3: Execute the batch
/task-batch .claude/plans/user-authentication.prq.md
# Or with options
/task-batch .claude/plans/user-authentication.prq.md --priority
Example Session
User: /create-task-batch "User profile feature"
Assistant: I'll help you create a task batch for "User profile feature".
[Phase 1: PROJECT CONTEXT]
Q: What is the main goal of this batch?
A: New Feature
Q: What is the project area?
A: Full Stack
Q: What is the target timeframe?
A: 1-2 days
[Phase 2: FEATURE DETAILS]
Q: Describe the main feature in detail:
A: Users should be able to view and edit their profile, upload avatar, and set preferences
Q: What are the MUST-HAVE components?
A: Authentication, Data validation, Testing
[Phase 3: TASK DECOMPOSITION]
Based on your input, I've decomposed the feature into these tasks:
1. [P1] Create user profile data model
2. [P1] Implement profile API endpoints (GET, PUT)
3. [P2] Create profile UI page
4. [P2] Add avatar upload functionality
5. [P3] Implement user preferences
6. [P3] Write unit and integration tests
[Phase 4: PRIORITY & DEPENDENCIES]
Q: How should tasks be prioritized?
A: Auto (by dependencies)
Q: Should execution stop on first failure?
A: No - Continue with remaining
[Phase 5: ACCEPTANCE CRITERIA]
Q: Acceptance criteria for "Create user profile data model":
A: Schema exists, migrations run, validation works
[OUTPUT]
Created: .claude/plans/user-profile-feature.prq.md
Ready to execute with:
/task-batch .claude/plans/user-profile-feature.prq.md
Anti-Patterns
- NEVER create tasks without acceptance criteria
- NEVER skip the dependency check
- NEVER create circular dependencies
- NEVER make tasks too large (should be 1-2 hours max)
- NEVER forget to validate the output before execution
Related Skills
/task-batch- Execute the created task list/clarify- Used internally for requirement gathering/orchestrator- Base orchestration workflow/retrospective- Post-batch analysis
Action Reporting (v2.93.0)
Esta skill genera reportes automáticos completos para trazabilidad:
Reporte Automático
Cuando esta skill completa, se genera automáticamente:
- En la conversación de Claude: Resultados visibles
- En el repositorio:
docs/actions/create-task-batch/{timestamp}.md - Metadatos JSON:
.claude/metadata/actions/create-task-batch/{timestamp}.json
Contenido del Reporte
Cada reporte incluye:
- ✅ Summary: Descripción de la tarea ejecutada
- ✅ Execution Details: Duración, iteraciones, archivos modificados
- ✅ Results: Errores encontrados, recomendaciones
- ✅ Next Steps: Próximas acciones sugeridas
Ver Reportes Anteriores
# Listar todos los reportes de esta skill
ls -lt docs/actions/create-task-batch/
# Ver el reporte más reciente
cat $(ls -t docs/actions/create-task-batch/*.md | head -1)
# Buscar reportes fallidos
grep -l "Status: FAILED" docs/actions/create-task-batch/*.md
Generación Manual (Opcional)
source .claude/lib/action-report-lib.sh
start_action_report "create-task-batch" "Task description"
# ... ejecución ...
complete_action_report "success" "Summary" "Recommendations"
Referencias del Sistema
-
Action Reports System - Documentación completa
-
action-report-lib.sh - Librería helper
-
action-report-generator.sh - Generador
Related skills
More from alfredolopez80/multi-agent-ralph-loop
stop-slop
A skill for removing AI-generated writing patterns ('slop') from prose. Eliminates telltale signs of AI writing like filler phrases, excessive hedging, overly formal language, and mechanical sentence structures. Use when: writing content that should sound human and natural, editing AI-generated drafts, cleaning up prose for publication, or any content that needs to sound authentic rather than AI-generated. Triggers: 'stop-slop', 'remove AI tells', 'clean up prose', 'make it sound human', 'edit AI writing'.
10iterate
Ralph Loop pattern with swarm mode: iterative execution until VERIFIED_DONE with multi-agent coordination. Use when: (1) iterative refinement needed, (2) quality gates must pass, (3) automated validation required. Triggers: /iterate, 'iterate until done', 'keep trying', 'fix until passing', 'loop until done'.
2gemini-cli
|
2minimax
Custom skill for minimax
1clarify
Intensive requirement clarification using structured AskUserQuestion workflow. Gathers MUST_HAVE (blocking) and NICE_TO_HAVE (optional) information before implementation. Use when: (1) starting new feature implementation, (2) requirements are ambiguous, (3) multiple approaches possible, (4) before writing any code. Triggers: /clarify, 'clarify requirements', 'ask questions', 'gather requirements'.
1security
Security audit with Codex + MiniMax second opinion. Integrates ralph-security agent (6 quality pillars, OWASP A01-A10). Uses LSP for code navigation during analysis. Use when: (1) /security is invoked, (2) task relates to security functionality.
1