Elephant-Goldfish Model
Helps your coding agent draft a feature spec, then has fresh reviewers poke holes in it
by Google
Spec Kit
by GitHubCLI toolWrite code
Adds spec-driven build, bug-fix and idea-check processes to your AI coding agent
About Spec Kit
Spec Kit adds ready-made processes to your AI coding agent: build a feature from a written spec, fix a bug with a recorded check, or assess whether an idea is worth building.
For DevelopersDevelopment teams
Made by GitHub
GitHub, Inc. publishes Spec Kit as an open source project under the MIT licence.
- Stars
- ★ 148k
- License
- MIT
- Cost
- Usage-based
- Platforms
- Mac · Linux · Win
- Level
- Intermediate
Overview
Spec Kit is an open source toolkit from GitHub that gives your AI coding agent structured processes, reusable templates and written records. You install the specify command, set up a project with specify init, then run the process commands in your agent's chat: write a spec, plan, split tasks and implement, or assess a bug or an idea. It suits developers who already use an AI coding agent.
Each process leaves written files, such as specs and verdicts, that you can read and edit between steps.
Try these first
Start with the first one: it is a small task you can check easily.
- /speckit-constitution Create principles focused on code quality, testing, and maintainability.First try · writes files
- /speckit-specify Build [describe the app in plain words].writes files
- /speckit-plan Use [tools you prefer]. Keep data local.writes files
- /speckit-taskswrites files
Example prompts written by Agent Flows Hub.
Features
From the creator's documentation.
| Feature | What it does | Runs |
|---|---|---|
| Spec-driven development | Carries a written spec through planning, tasks, implementation and a convergence check | Your computer |
| Bug fixing | Optional extension that assesses a cause, fixes it and records a verification verdict | Your computer |
| Idea assessment | Optional extension that gathers evidence and ends in a go, clarify or stop decision | Your computer |
| Agent integrations | Connects to many coding agents; add or switch agents in a project | Your computer |
| Extensions and presets | Add commands, quality gates or team standards, including from community catalogs | Online catalogs |
Good to know
You build software with an AI coding agent and want a written spec, plan and task list before it writes code. Spec Kit adds those steps as commands in your agent, and its extensions cover bug fixing and checking whether an idea is worth building.
Limits
- The process commands run inside your AI coding agent's chat, not in the terminal, and need Python 3.11+, uv and a supported agent.
- Community catalog listings are not audits or endorsements, so read community source code before installing it.
Notes by Agent Flows Hub (original project name: Spec Kit), written from the creator's documentation, not from running it ourselves. Spec Kit README · Spec Kit installation guide
Stuck setting it up?
Tell us what happened. We take a look first; if it needs paid work, you get an estimate before anything starts.
More to write code
Other cli tools
- Agent Reach Lets your AI agent search and read YouTube, Reddit, X, Bilibili, Xiaohongshu and more
- agent-desktop Lets your AI agent read and click through the desktop apps on your computer
- agent-device Lets your coding agent tap through a mobile or desktop app and save screenshots as proof
- ai-memory Gives your coding agents a shared memory so the next one picks up where you stopped
- ccusage Shows how many tokens your coding agents used and what that would cost, day by day
- Claude Code Templates Installs ready-made skills, agents, commands and MCP setups into Claude Code in one step