SpecKit Workflows

SpecKit is a suite of AI-powered workflows that manage the entire feature development lifecycle — from a natural language description to a fully implemented and analyzed feature.

The Feature Lifecycle

Natural Language Idea
        ↓
  /speckit.specify    ← Create feature specification
        ↓
  /speckit.clarify    ← Resolve ambiguities (optional)
        ↓
  /speckit.plan       ← Generate implementation plan
        ↓
  /speckit.tasks      ← Break into actionable task list
        ↓
  /speckit.implement  ← Execute tasks one by one
        ↓
  /speckit.analyze    ← Verify cross-artifact consistency

Available Workflows

/speckit.specify

Purpose: Transform a natural language feature description into a structured specification document (spec.md).

Output: A comprehensive spec.md covering user stories, acceptance criteria, and technical requirements.

Usage:

/speckit.specify

I want to add user authentication with email and password login,
with JWT tokens and refresh token rotation.

---

/speckit.clarify

Purpose: Identify underspecified areas in the current feature spec. Asks up to 5 targeted clarification questions and encodes answers back into the spec.

When to use: After /speckit.specify when the spec has ambiguous areas before planning.

---

/speckit.plan

Purpose: Execute implementation planning based on the spec. Generates design artifacts including architecture diagrams, API contracts, and database schemas.

Output: A plan.md with the technical approach, proposed file changes, and dependencies.

---

/speckit.tasks

Purpose: Generate an actionable, dependency-ordered task list from the available design artifacts.

Output: A tasks.md with numbered, granular tasks including file paths, acceptance criteria, and implementation notes.

---

/speckit.implement

Purpose: Execute the implementation plan by processing all tasks defined in tasks.md sequentially. Each task is implemented and committed atomically.

Key behavior: Implements one task → commits → proceeds to next task. Never batches.

---

/speckit.analyze

Purpose: Perform a non-destructive cross-artifact consistency and quality analysis across spec.md, plan.md, and tasks.md after task generation.

Checks for: Gaps, contradictions, missing requirements, and alignment issues between artifacts.

---

/speckit.checklist

Purpose: Generate a custom implementation checklist for the current feature based on the spec.

---

/speckit.taskstoissues

Purpose: Convert tasks from tasks.md into actionable, dependency-ordered GitHub Issues.

---

/snowdreamtech.init

Purpose: Initialize the development environment for this project. Installs all required tools, configures pre-commit hooks, and sets up language-specific linters.

Running Workflows

Workflows are available in all supported AI IDEs via the slash command syntax. The exact invocation depends on the IDE:

IDE	Syntax
Cursor	/speckit.specify
Windsurf	/speckit.specify
Cline	/speckit.specify
Claude	@[/speckit.specify]
Gemini	/speckit.specify

Customizing Workflows

Workflow definitions live in .agent/workflows/. Each file is a Markdown document with ---description--- frontmatter and step-by-step instructions.

To create a custom workflow:

1. Add a new .md file to .agent/workflows/
2. Follow the existing format with YAML frontmatter
3. The workflow automatically becomes available in all supported IDEs