insightflo

tasks-init

16
2
# Install this skill:
npx skills add insightflo/claude-impl-tools --skill "tasks-init"

Install specific skill from multi-skill repository

# Description

Interactively scaffolds a TASKS.md file from scratch. Use this at project start, whenever TASKS.md is missing, or whenever task structure is needed. Invoke immediately on "create tasks", "generate TASKS.md", "start project", or "organize to-dos" requests. Runs standalone. Triggered by /tasks-init.

# SKILL.md


name: tasks-init
description: Interactively scaffolds a TASKS.md file from scratch. Use this at project start, whenever TASKS.md is missing, or whenever task structure is needed. Invoke immediately on "create tasks", "generate TASKS.md", "start project", or "organize to-dos" requests. Runs standalone. Triggered by /tasks-init.
triggers:
- /tasks-init
- νƒœμŠ€ν¬ μ΄ˆκΈ°ν™”
- TASKS λ§Œλ“€μ–΄μ€˜
- νƒœμŠ€ν¬ 생성
- ν”„λ‘œμ νŠΈ μ‹œμž‘
- 할일 정리
version: 2.1.0


Tasks Init (Standalone)

A lightweight skill that interactively creates a TASKS.md file.
Fully standalone β€” runs independently with no external dependencies.

Role

  • Collects project information interactively
  • Generates detailed tasks via Specialist context injection
  • Auto-detects dependencies and adds metadata
  • Produces a domain-guarded TASKS.md (backend/frontend separated)

v2.0.0 update: Dependency-aware, Domain-guarded, Specialist integration

Execution Flow

/tasks-init
     ↓
β”Œβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”
β”‚ Stage 1: Collect project info (AskUserQuestion)             β”‚
β”‚   β€’ Project name                                            β”‚
β”‚   β€’ Key features (3–5)                                      β”‚
β”‚   β€’ Tech stack (auto-detected)                              β”‚
β””β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”˜
     ↓
β”Œβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”
β”‚ Stage 2: Analyze existing code (automatic)                  β”‚
β”‚   β€’ Parse package.json / pyproject.toml                     β”‚
β”‚   β€’ Scan directory structure (domain detection)             β”‚
β”‚   β€’ Analyze import/require dependencies                     β”‚
β”‚   β€’ Collect existing TODO markers                           β”‚
β””β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”˜
     ↓
β”Œβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”
β”‚ Stage 3: Specialist context injection (v2.0 NEW)            β”‚
β”‚   β€’ Backend Specialist β†’ detail backend tasks               β”‚
β”‚   β€’ Frontend Specialist β†’ detail frontend tasks             β”‚
β”‚   β€’ Security Specialist β†’ add security-related tasks        β”‚
β””β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”˜
     ↓
β”Œβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”
β”‚ Stage 4: Generate TASKS.md (Dependency-aware)               β”‚
β”‚   β€’ Auto-calculate dependencies (deps field)                β”‚
β”‚   β€’ Separate by domain (domain field)                       β”‚
β”‚   β€’ Auto-classify risk (risk field)                         β”‚
β”‚   β€’ Detect file conflicts (files field)                     β”‚
β””β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”˜
     ↓
β”Œβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”
β”‚ Stage 5: User confirmation + next-step guidance             β”‚
β”‚   β†’ Confirm owner-based auto-routing + recommend            β”‚
β”‚     /agile auto or /team-orchestrate                        β”‚
β””β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”˜

Stage 1: Collect Project Info

{
  "questions": [
    {
      "question": "What are the main features of your project? (e.g., user auth, product catalog, payments)",
      "header": "Key Features",
      "options": [
        {"label": "Type manually", "description": "Write the feature list yourself"}
      ],
      "multiSelect": false
    }
  ]
}

Stage 2: Code Analysis

# Detect tech stack
ls package.json pyproject.toml requirements.txt Cargo.toml go.mod 2>/dev/null

# Directory structure
ls -d */ 2>/dev/null | head -10

# Collect existing TODOs
grep -rn "TODO\|FIXME\|XXX" --include="*.ts" --include="*.tsx" --include="*.py" 2>/dev/null | head -20

Stage 3: TASKS.md Template

# TASKS.md

> Created: {date}
> Project: {project_name}

---

## T0 - Skeleton (Structure)

- [ ] T0.1: Initial project setup
- [ ] T0.2: Create directory structure
- [ ] T0.3: Configure routing/navigation
- [ ] T0.4: Define dummy data structures

## T1 - Muscles (Core Features)

{Auto-generated tasks per feature}

- [ ] T1.1: {feature1} backend implementation
- [ ] T1.2: {feature1} frontend implementation
- [ ] T1.3: {feature2} backend implementation
- [ ] T1.4: {feature2} frontend implementation

## T2 - Muscles Advanced (Advanced Features)

- [ ] T2.1: Error handling
- [ ] T2.2: Loading state management
- [ ] T2.3: Caching layer

## T3 - Skin (Polish)

- [ ] T3.1: Apply design system
- [ ] T3.2: Responsive layout
- [ ] T3.3: Animations/transitions
- [ ] T3.4: Accessibility review

Stage 4: Next-Step Guidance

{
  "questions": [
    {
      "question": "TASKS.md has been created. What would you like to do next?",
      "header": "Next Step",
      "options": [
        {"label": "Start implementation (/agile auto)", "description": "Layer-by-layer auto implementation for ≀30 tasks"},
        {"label": "Parallel orchestration (/team-orchestrate)", "description": "Dependency-based parallel execution for 30–80 tasks"},
        {"label": "Manual", "description": "Edit tasks manually and proceed at your own pace"}
      ],
      "multiSelect": false
    }
  ]
}

In the generated TASKS.md, owner determines the default executor. Set model only when you need to override the owner/model-routing auto-routing.

Skill Relationship
/tasks-migrate Consolidate existing legacy files
/agile auto Execute the generated TASKS.md (≀30 tasks)
/team-orchestrate Parallel orchestration (30–80 tasks)
/governance-setup Planning for large-scale projects

Last Updated: 2026-03-03 (v2.0.0)

File Structure

skills/tasks-init/
β”œβ”€β”€ SKILL.md                    # Skill definition
β”œβ”€β”€ scripts/
β”‚   β”œβ”€β”€ analyze.js              # Code analysis (tech stack, dependencies, TODOs)
β”‚   β”œβ”€β”€ generate.js             # Task generation (Specialist context injection)
β”‚   └── tasks-init.sh           # Main entry point
└── templates/
    β”œβ”€β”€ task-metadata.yaml      # Metadata format description
    └── TASKS.md                # Template for the generated TASKS.md

Usage

Run directly via CLI

# Default usage (creates TASKS.md in the current directory)
cd skills/tasks-init/scripts
./tasks-init.sh

# Specify output file
./tasks-init.sh --output ../TASKS.md

# Specify feature list
./tasks-init.sh --features "user-auth,product-catalog,payment"

Run as a skill

/tasks-init

Claude will interactively collect project information and then generate TASKS.md.

# Supported AI Coding Agents

This skill is compatible with the SKILL.md standard and works with all major AI coding agents:

Learn more about the SKILL.md standard and how to use these skills with your preferred AI coding agent.