站内导读
定位:Spec Kit 是 GitHub 开源的工具包,为 AI 编程代理补上结构化流程、可复用模板和可留档的产出,用在三类场景:从零构建功能或应用、诊断并修复缺陷、判断一个想法是否值得继续投入。
核心能力
- 预设三条彼此独立的入口:规格驱动开发、缺陷修复、想法评估。
- 作业按工具包定义的结构与模板推进,代理的产出保持有序、便于事后复核。
- 每次运行的产出都会留档为文档,方便回看与共享。
位置づけ:Spec Kit は GitHub がオープンソースで公開するツールキットで、AI コーディングエージェントに構造化されたプロセス、再利用可能なテンプレート、保存できる成果物を与えます。対象は 3 つの場面です。機能やアプリをゼロから構築すること、欠陥を診断して修正すること、アイデアに続ける価値があるか判断することです。
主な機能
- 仕様駆動開発、欠陥修正、アイデア評価という、互いに独立した 3 つの入口が用意されています。
- 作業はツールキットが定める構造とテンプレートに沿って進むため、エージェントの成果物が整理され、後から確認しやすくなります。
- 実行ごとの成果物はドキュメントとして残り、見返したり共有したりできます。
정의: Spec Kit은 GitHub가 오픈소스로 공개한 툴킷으로, AI 코딩 에이전트에 구조화된 프로세스, 재사용 가능한 템플릿, 보관할 수 있는 산출물을 제공합니다. 사용 장면은 세 가지입니다. 기능이나 앱을 처음부터 구축하기, 결함을 진단하고 수정하기, 아이디어를 계속 추진할 가치가 있는지 판단하기입니다.
핵심 기능
- 사양 주도 개발, 결함 수정, 아이디어 평가라는 서로 독립적인 세 가지 진입점이 미리 마련되어 있습니다.
- 작업이 툴킷이 정한 구조와 템플릿을 따라 진행되므로 에이전트의 산출물이 정리되고 나중에 검토하기 쉽습니다.
- 실행마다 산출물이 문서로 남아 다시 살펴보거나 공유할 수 있습니다.
What it is: Spec Kit is GitHub's open-source toolkit that gives AI coding agents structured processes, reusable templates and archivable outputs. It serves three scenarios: building a feature or application from scratch, diagnosing and fixing defects, and judging whether an idea is worth pursuing.
Key capabilities
- Three independent entry points are provided: spec-driven development, defect fixing, and idea assessment.
- Work follows the structure and templates the toolkit defines, so agent output stays organized and reviewable.
- The outputs of each run are kept as documents, making them easy to revisit and share.
README · 详细文档
来源 · 查看原文
🌱 Spec Kit
Build with a spec, fix a bug, or assess an idea — with your coding agent.
Spec Kit is an open source toolkit that gives AI coding agents structured processes, reusable templates, and documented outcomes. Start with a ready-made process, customize it, or bring your own.
Choose your process
| What you need | Process | Outcome |
|---|---|---|
| Build a feature or application | Spec-Driven Development | A specification carried through planning, implementation, and convergence |
| Diagnose and repair broken behavior | Bug fixing | An assessed cause, scoped fix, and recorded verification |
| Decide whether an idea deserves investment | Idea assessment | An evidence-backed go, clarify, or stop decision |
| Adopt or create a different process | Your own process | A process adapted or assembled from reusable primitives |
These are independent entry points, not mandatory phases. SDD ships in core; bug fixing and assessment are bundled extensions you install when needed, and the primitives below let you adopt or create other processes.
Capabilities
Spec Kit combines ready-to-use processes with composable mechanisms:
| Capability | Spec Kit mechanism |
|---|---|
| Connect to or switch between coding agents | Integration |
| Add a command, quality gate, or complete process | Extension |
| Connect a process to an external tool or service | Extension |
| React to coding-agent lifecycle events | Extension runtime event |
| Run extension commands before or after Spec Kit commands | Extension hook |
| Adapt templates, commands, terminology, or methodology | Preset |
| Enforce organizational or regulatory standards | Preset |
| Override a template for one project | Project-local override |
| Automate prompts, commands, gates, and shell tasks | Workflow |
| Modify an installed workflow without forking it | Workflow overlay |
| Add reusable executable behavior to workflows | Workflow step |
| Install a curated setup for a role or team | Bundle |
| Discover community components | Catalog |
| Host a private catalog of approved components | Catalog |
| Access private catalogs and component downloads | Authentication |
| Work offline or behind a firewall | Local sources and self-hosted catalogs |
| Inspect which layer supplies a command, template, or script | Artifact inventory |
Search the active catalogs with commands such as specify extension search, specify preset search, specify workflow search, and specify bundle search. Spec Kit includes community catalogs, and organizations can add their own curated sources. Catalog listings are not audits or endorsements, so review community source code before installing or running it.
Get started
You need Python 3.11+, uv, and a supported AI coding agent on Linux, macOS, or Windows.
Run these commands in your terminal to install Spec Kit and create a project. This example uses GitHub Copilot.
uv tool install specify-cli
specify init my-project --integration copilot
cd my-project
After installing the Specify CLI and creating your project, launch your coding agent in the project directory, then choose a process below.
For other setups:
- Installation for pinned releases, other installers, CI, and troubleshooting.
- Existing projects to start with code you already have.
- Other agents for integration keys and invocation syntax.
- Upgrade to update an existing installation.
Spec-Driven Development
Define what to build and why, then turn the specification into a plan, tasks, and working code.
The minimal flow is shown below. Create a constitution once per project and repeat the remaining steps for each feature. Run these skills in your agent's chat, not your terminal, one at a time, reviewing each result before continuing.
This example uses GitHub Copilot's default skills syntax. For another agent, use its equivalent invocation syntax.
/speckit-constitution Create principles focused on code quality, testing, and maintainability.
/speckit-specify Build a photo organizer with albums grouped by date and a tile preview of each album.
/speckit-plan Use Vite with vanilla JavaScript. Keep images local and store metadata in SQLite.
/speckit-tasks
/speckit-implement
/speckit-converge
Repeat implement and converge until convergence reports Converged. For clarification, checklists, and consistency analysis, see the SDD walkthrough.
SDD walkthrough · Command reference
Bug fixing
Assess the cause, fix the bug, and verify the original symptom is resolved. No SDD feature workflow is required first.
Install the bug-fixing extension in your terminal, from the project directory:
specify extension add bug
Run these skills in your agent's chat, not your terminal, one at a time, reviewing each result before continuing.
This example uses GitHub Copilot's default skills syntax. For another agent, use its equivalent invocation syntax.
/speckit-bug-assess "Submitting an empty password crashes the login form." slug=login-crash
/speckit-bug-fix slug=login-crash
/speckit-bug-test slug=login-crash
Review the reports in .specify/bugs/login-crash/. The final verdict is verified, partial, or failed; only verified confirms a successful fix.
Bug-fixing walkthrough · Command reference
Idea assessment
Gather evidence to decide whether an idea is worth pursuing. No source code or SDD feature workflow is required.
Install the assessment extension in your terminal, from the project directory:
specify extension add assess
Run these skills in your agent's chat, not your terminal, one at a time, reviewing each result before continuing.
This example uses GitHub Copilot's default skills syntax. For another agent, use its equivalent invocation syntax.
/speckit-assess-intake "Let users work offline and sync when they reconnect." slug=offline-mode
/speckit-assess-research slug=offline-mode
/speckit-assess-define slug=offline-mode
/speckit-assess-shape slug=offline-mode
/speckit-assess-decide slug=offline-mode
Review the artifacts in .specify/assessments/offline-mode/ and the final decision: go, needs-clarification, or kill. Resolve unknowns by editing the existing Markdown artifacts, directly or with your agent, rather than rerunning whole stages. If you choose to build a software idea with a go decision, establish a project constitution if needed, then pass the handoff summary from decision.md to /speckit-specify. Stopping with a documented reason is also a useful outcome.
Assessment walkthrough · Command reference
Customize or bring your own process
Customize Spec-Driven Development, bug fixing, or idea assessment to fit your team's standards. Or bring your own process without adopting any of them. Combine the capabilities above to match your goal:
- Adapt an existing process with a preset, such as requiring regulatory traceability in plans. Use a project-local override for a one-off template change.
- Bring your own process by adding its commands through an extension and exposing them to your coding agent through an integration.
- Automate commands and checks in a workflow. Use overlays to adapt an installed workflow without forking it, or add custom workflow steps when the built-in types are not enough.
- Share a reusable setup by packaging its extensions, presets, workflows, and workflow steps in a bundle.
Start with the smallest change that meets your needs; you do not have to replace the whole process.
Customization guide · Community components and walkthroughs
Documentation
- CLI reference and process commands
- How-to guides for applying Spec Kit to your project work
- SDD philosophy, full methodology, and evolving existing specs
- Project history
- How Spec Kit uses Spec Kit
In the news
- Open Source Friday: Spec Kit Part 2 (October 16, 2026) — Manfred Riem and Nicole Haugen discuss Spec Kit 1.0 and the expanded toolkit.
- Living Specs in SpecKit (October 15, 2026) — Alfredo Perez explores maintaining capability-level specifications as a project evolves.
- Spec Kit Customization Series (October 8–10, 2026) — Jesse Liberty walks through presets and extension authoring, including testing and publishing.
- From Spec-First to Enterprise-Ready: Extending GitHub Spec Kit (October 7, 2026) — Apoorv Gupta explains enterprise customization through presets, extensions, bundles, workflows, and governed catalogs.
- Open Source Friday with Spec Kit (May 8, 2026) — An introduction and SDD demonstration with Manfred Riem.
Star history
Thanks to everyone who helps Spec Kit shine.
Support and contributing
Report a bug or request a feature · Contributing guide · Code of conduct
Spec Kit is MIT licensed.