Skip to main content
Spec-Kit is a prompt library for specification-driven development. It helps you analyze requirements, create detailed specs, and generate implementation checklists that can be automated with Auto Run.

​
What is Spec-Kit?

Spec-Kit provides a structured workflow for breaking down complex features into actionable tasks:
  1. Analyze - Understand the problem space
  2. Specify - Create detailed specifications
  3. Plan - Generate implementation tasks
  4. Checklist - Convert to Auto Run format
  5. Implement - Execute tasks with AI assistance

​
Available Commands

Spec-Kit commands are accessed via slash commands in the AI terminal:

​
/speckit:help

Show available Spec-Kit commands and usage examples. 
/speckit:help

​
/speckit:analyze

Analyze a feature request or problem to understand requirements, constraints, and edge cases. 
/speckit:analyze Add user authentication to the dashboard
What it does:
  • Extracts core requirements
  • Identifies edge cases and error scenarios
  • Lists technical constraints
  • Suggests alternatives and tradeoffs

​
/speckit:clarify

Ask clarifying questions about ambiguous requirements. 
/speckit:clarify The search feature should be fast
What it does:
  • Identifies unclear or ambiguous requirements
  • Generates specific questions to resolve ambiguity
  • Suggests concrete acceptance criteria

​
/speckit:specify

Generate a detailed technical specification document. 
/speckit:specify Implement real-time notifications
Output format:
  • Problem statement
  • Requirements (functional and non-functional)
  • Technical design
  • API contracts
  • Database schema changes
  • Security considerations
  • Testing strategy

​
/speckit:plan

Create a high-level implementation plan with milestones. 
/speckit:plan Migrate to new authentication system
What it does:
  • Breaks work into phases
  • Identifies dependencies
  • Estimates complexity
  • Suggests rollout strategy

​
/speckit:tasks

Generate granular implementation tasks from a specification. 
/speckit:tasks [paste your specification here]
Output format:
  • Numbered task list
  • Each task is atomic and testable
  • Tasks are ordered by dependency
  • Includes testing and documentation tasks

​
/speckit:checklist

Convert tasks into Auto Run checklist format. 
/speckit:checklist [paste task list here]
Output: Markdown checklist ready for Auto Run: 
## Implementation Checklist

- [ ] Task 1: Set up authentication routes
- [ ] Task 2: Create user model
- [ ] Task 3: Implement password hashing

​
/speckit:taskstoissues

Convert task list to GitHub/GitLab issue format. 
/speckit:taskstoissues [paste task list here]

​
/speckit:implement

Generate implementation code for a specific task. 
/speckit:implement Task: Add JWT token validation middleware

​
/speckit:constitution

Review code against project coding standards and best practices. 
/speckit:constitution [paste code here]

​
Workflow Example

1

Analyze the requirement

/speckit:analyze Add real-time collaboration to the document editor
Review the analysis output to understand scope and complexity.
2

Clarify ambiguities

/speckit:clarify What does 'real-time' mean in terms of latency?
Answer the questions to refine requirements.
3

Generate specification

/speckit:specify Implement WebSocket-based real-time collaboration
with operational transformation for conflict resolution
Save the specification document to a file.
4

Create implementation plan

/speckit:plan [paste specification]
Review milestones and adjust scope if needed.
5

Generate task list

/speckit:tasks [paste specification]
Get a detailed, ordered list of implementation tasks.
6

Convert to checklist

/speckit:checklist [paste task list]
Save the output as a markdown file (e.g., collab-tasks.md).
7

Run with Auto Run

Enable Auto Run in the right panel, select your checklist file, and start execution.

​
Integration with Auto Run

Spec-Kit is designed to work seamlessly with Auto Run:
  1. Use /speckit:checklist to generate Auto Run-compatible markdown
  2. Save the checklist to a .md file in your workspace
  3. Select it in the Auto Run panel
  4. Each task becomes a separate AI session with clean context
Best Practice: Keep tasks atomic (30-60 minutes each) for optimal Auto Run performance.

​
Customization

Spec-Kit prompts are stored in src/prompts/speckit/ and can be customized:
  • speckit.analyze.md - Analysis template
  • speckit.specify.md - Specification format
  • speckit.checklist.md - Checklist conversion
  • speckit.implement.md - Implementation guidance

​
Best Practices

Always run /speckit:analyze first to understand the full scope before diving into implementation.
Use /speckit:clarify to resolve ambiguities early. It’s faster to ask questions than to refactor later.
Break large features into multiple smaller specs. Each spec should fit in a single Auto Run session.
AI-generated task lists are starting points. Review and adjust based on your codebase and constraints.
Run /speckit:tasks multiple times with refined specifications to get better task breakdowns.

OpenSpec Commands

Complementary specification workflow

Auto Run & Playbooks

Execute checklists automatically

Build docs developers (and LLMs) love

Get started for freeTalk to us