nathangeerinck/dotfiles/spec-writer

Claude Code and agent skill by nathangeerinck/dotfiles

Trust tier 3
  • community
Source check pending

Checking that the SKILL.md file is still in the repository.

Create detailed software specifications through iterative Q&A brainstorming. Use when starting a new project, feature, or app. Produces spec.md, architecture docs, implementation plans with step-by-step prompts, and todo checklists. Triggers include "write a spec", "plan this project", "brainstorm this idea", "create implementation plan", or any new project/feature planning.

Install

npx vskill@latest install nathangeerinck/dotfiles/spec-writer

SKILL.md

Spec Writer

Transform ideas into developer-ready specifications through structured brainstorming and planning.

The Process

Phase 1: Idea Honing (Brainstorm)

Ask ONE question at a time to develop a thorough spec. Each question builds on previous answers. Goal: detailed specification ready for implementation.

Start with:

I'll help you develop a detailed spec for this idea. I'll ask one question at a time.
Let's start: [First clarifying question about the core problem/goal]

Key areas to explore:

  • Core problem: What pain point does this solve?
  • Users: Who uses this? What are their workflows?
  • Features: What are the must-haves vs nice-to-haves?
  • Data: What entities exist? How do they relate?
  • Integrations: What external systems are involved?
  • Constraints: Performance, security, tech stack requirements?
  • Edge cases: What could go wrong? How to handle it?

When brainstorming concludes naturally, compile into SPEC.md.

Phase 2: Architecture

Create ARCHITECTURE.md covering:

  • System overview diagram (Mermaid)
  • Component breakdown
  • Data flow
  • Key technical decisions
  • File/folder structure
Phase 3: Implementation Plan

Break the spec into small, iterative chunks:

  1. First pass: Major phases/milestones
  2. Second pass: Break each phase into tasks
  3. Third pass: Ensure tasks are:
  • Small enough for safe implementation with tests
  • Big enough to move the project forward
  • Building on previous tasks (no orphaned code)
  • Ending with integration/wiring

Output: PROMPT_PLAN.md with ready-to-use prompts for each step.

Phase 4: Checklist

Generate TODO.md as a checkable list that tracks progress.

Output Structure

analysis/ # or docs/, specs/, etc.
├── SPEC.md # Full specification
├── ARCHITECTURE.md # System design
├── PROMPT_PLAN.md # Step-by-step implementation prompts
├── TODO.md # Checklist version
└── phases/ # Optional: detailed phase docs
├── 01-phase-name.md
├── 02-phase-name.md
└── ...

SPEC.md Template

markdown
# [Project Name] Specification
## Overview
[One paragraph summary]
## Problem Statement
[What pain point this solves]
## Users & Personas
[Who uses this and how]
## Features
### Must Have (MVP)
-[ ] Feature 1: [description]
-[ ] Feature 2: [description]
### Should Have
-[ ] Feature 3: [description]
### Nice to Have
-[ ] Feature 4: [description]
## Data Model
[Entities, relationships, key fields]
## User Flows
[Step-by-step workflows for key actions]
## Technical Requirements
-Stack: [technologies]
-Integrations: [external services]
-Performance: [requirements]
-Security: [considerations]
## Edge Cases & Error Handling
[What could go wrong and how to handle it]
## Success Criteria
[How do we know it's done? What tests prove it works?]
## Open Questions
[Anything still TBD]

PROMPT_PLAN.md Format

Each prompt section should:

  1. Reference what exists (files, patterns to follow)
  2. State the specific task
  3. Include verification criteria (tests, expected output)
  4. Be tagged with markdown code blocks for easy copy/paste
markdown
## Phase 1: [Name]
### Step 1.1: [Task Name]
**Context**: [What exists, what patterns to follow]
**Prompt**:
\`\`\`
[Ready-to-use prompt for the AI agent]
Verification:
-[ ] Test X passes
-[ ] Feature Y works as expected
\`\`\`
### Step 1.2: [Next Task]
...

Best Practices

  1. One question at a time during brainstorming — don't overwhelm
  2. Be specific — vague specs produce vague implementations
  3. Include verification — every feature needs a "done" signal
  4. Reference patterns — point to existing code as examples
  5. Small steps — each prompt should be achievable in one session
  6. No orphans — every step ends with integration
  7. Test-first thinking — define expected behavior before implementation

When to Use Each Phase

SituationStart At
Vague idea, exploringPhase 1 (Brainstorm)
Clear requirements, need structurePhase 2 (Architecture)
Architecture known, need execution planPhase 3 (Plan)
Resuming existing projectPhase 4 (TODO review)

Security

  • Security scanPassed, 90 of 100Pattern checks on the skill's files.
  • Model reviewPassed

Quality evaluation

  • Task evaluationNot runNo evaluation yet. Run one in Skill Studio.

Source

Popularity

7-day trend score
0

Not refreshed yet.

Works with

Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.

  • Claude Code
  • Cursor
  • GitHub Copilot
  • Windsurf
  • Codex
  • Gemini CLI
  • Cline
  • Amp
  • Roo Code
  • Goose
  • Continue
  • Kilo Code
  • Junie
  • Trae
  • OpenHands
  • Qwen Code
  • AI Agent
  • Tabnine Chat
  • Sourcegraph Cody
  • Amazon Q
  • Firebase Genkit
  • Sweep
  • Devin
  • Phind
  • Mentat
  • Claude Desktop
  • Aider
  • GPT Pilot
  • Plandex
  • bolt.new
  • v0
  • Replit Agent
  • Anse
  • TypingMind
  • BetterChatGPT
  • LibreChat
  • Warp AI
  • Pieces
  • Fig AI
Embed badge
vskill VERIFIED badge
[![vskill VERIFIED](https://verified-skill.com/api/v1/skills/nathangeerinck/dotfiles/spec-writer/badge)](https://verified-skill.com/skills/nathangeerinck/dotfiles/spec-writer)

Something wrong with this skill or its page? Tell us.