Repolode
gotalab

gotalab/cc-sdd

Installs a spec-driven development workflow as Agent Skills for AI coding agents like Claude Code, Codex, and Cursor, covering requirements, design, tasks, and autonomous implementation.

Open-source alternative to:Kiro

Stars
3,725
Forks
285
Open issues
19
Last commit
8 days ago

What is it?

cc-sdd is an installer that adds a spec-driven software development workflow to AI coding agents. Running one command sets up 17 Agent Skills that guide the agent through discovery, requirements, design, task breakdown, and implementation. Specs are written as markdown files such as requirements.md, design.md, and tasks.md, and humans approve them at phase gates before code is written.

The project supports eight coding agents, including Claude Code and Codex (marked stable) and Cursor, GitHub Copilot, OpenCode, Gemini CLI, and Antigravity (marked beta). Its /kiro-impl skill runs approved tasks one at a time. On hosts with native subagents, each task gets a fresh implementer using TDD and an independent reviewer; otherwise the work runs inline in the main context. The workflow is designed to be compatible with specs written for Kiro IDE.

Key features

  • 01Installs 17 Agent Skills for each supported coding agent with one npx command.
  • 02Covers discovery, requirements in EARS format, design with a file structure plan, and task breakdown.
  • 03Runs approved tasks with TDD and independent review when native subagents are available.
  • 04Tracks task boundaries and dependencies through annotations in tasks.md.
  • 05Supports 14 languages for generated documentation.

Use cases

  • Turning a product idea into requirements, design, and tasks before an agent writes code.
  • Running a larger approved task list with an agent while keeping human approval gates.
  • Extending an existing codebase by first capturing its boundaries in steering and spec files.
  • Splitting a large initiative into multiple specs ordered by dependency.

Quick start

  1. 1

    Change into your project folder

    cd your-project
  2. 2

    Install the default Claude Code Skills

    npx cc-sdd@latest
  3. 3

    In your agent, start discovery with your idea

    /kiro-discovery <idea>

Things to know

  • Cursor, Copilot, Devin, OpenCode, Gemini CLI and Antigravity support is in beta.
  • Autonomous implementation with per-task review depends on native subagents being available in your host.
  • Legacy modes are deprecated and the Windsurf and Codex legacy options are blocked or deprecated.

Topics

FAQ

What is cc-sdd?
Installs a spec-driven development workflow as Agent Skills for AI coding agents like Claude Code, Codex, and Cursor, covering requirements, design, tasks, and autonomous implementation. cc-sdd is an installer that adds a spec-driven software development workflow to AI coding agents. Running one command sets up 17 Agent Skills that guide the agent through discovery, requirements, design, task breakdown, and implementation. Specs are written as markdown files such as requirements.md, design.md, and tasks.md, and humans approve them at phase gates before code is written.
Is cc-sdd free to use? What license does it have?
cc-sdd is open source under the MIT license. Check the license terms before using it in commercial products.
Is cc-sdd still maintained?
Actively maintained. The last commit to cc-sdd was on Oct 13, 2026.
What can I use cc-sdd for?
Turning a product idea into requirements, design, and tasks before an agent writes code. Running a larger approved task list with an agent while keeping human approval gates. Extending an existing codebase by first capturing its boundaries in steering and spec files. Splitting a large initiative into multiple specs ordered by dependency.
How do I install cc-sdd?
Change into your project folder: cd your-project Install the default Claude Code Skills: npx cc-sdd@latest In your agent, start discovery with your idea: /kiro-discovery <idea>
Where can I get cc-sdd?
The source code and installation instructions are on GitHub: https://github.com/gotalab/cc-sdd

Summary generated by AI from the project's README. GitHub stats refresh daily.

Related projects