Spec Kit
GitHub's open-source toolkit for spec-driven development with any AI coding agent
About
Spec Kit is GitHub's open-source toolkit for Spec-Driven Development (SDD): instead of prompting an AI agent ad hoc, you align on intent first — writing a specification, a technical plan, and a task breakdown — and the agent then implements against those artifacts. A 'specify' CLI scaffolds a '.specify/' directory and per-agent skills (e.g. /speckit.constitution, /speckit.specify, /speckit.plan, /speckit.tasks, /speckit.implement) into your project, which you invoke in your agent's chat one step at a time.
Three independent entry points ship: Spec-Driven Development, bug fixing, and idea assessment. It supports 38+ AI coding agents (Copilot, Claude Code, Cursor, Codex, Gemini CLI, Kilo Code, MiniMax Code, ...) and works with existing codebases, not just greenfield projects. GitHub describes it as an experimental initiative and community feedback notes higher token consumption from detailed specs — a real trade-off against the structure it buys.
What it can do
- • Three independent processes: Spec-Driven Development, bug fixing, and idea assessment (SDD ships in core; the others are bundled extensions)
- • Generates structured Markdown artifacts (.specify/) plus per-agent /speckit-* skills that act as explicit contracts between human intent and agent execution
- • 38+ agent integrations: GitHub Copilot, Claude Code, Cursor, Codex, Gemini CLI, Kilo Code, MiniMax Code, Docker Agent, and more
- • Growing extension ecosystem incl. a bundled GitHub extension (taskstoissues), OpenSpec, OWASP LLM Threat Model, and community catalog extensions
- • Brownfield support: initialize into existing repos with --here / --force; automation scripts in Bash, PowerShell, and Python variants
- • Pinned releases, one-time usage via uvx, air-gapped install support, and 'specify check' tool verification
Setup
- Install Python 3.11+ and uv ('curl -LsSf https://astral.sh/uv/install.sh | sh').
- Install the CLI: 'uv tool install specify-cli' (official PyPI package; alternatives: 'pipx install specify-cli', or pin a release with 'uv tool install specify-cli --from git+https://github.com/github/spec-kit.git@vX.Y.Z').
- Initialize a project: 'specify init my-project --integration claude' (replace 'claude' with your agent key: copilot, cursor, codex, gemini, ...; interactive prompts unless '--non-interactive').
- Launch your coding agent in the project directory, then run the generated '/speckit-*' skills in chat one at a time and review each result: '/speckit.constitution' -> '/speckit.specify' -> '/speckit.plan' -> '/speckit.tasks' -> '/speckit.implement'. These are agent skills, not terminal commands.
- 'specify check' verifies required tools; 'specify self upgrade' updates the CLI.
Prerequisites
Python 3.11+; uv (or pipx); Git; a supported AI coding agent (38+ integrations); an internet connection for package install.
Use cases
Works with
Trust score — 9/10
Scored 9/10 (verified tier, metrics-v1). Evidence: 140,964 GitHub stars (log-scaled vs catalog peers); blended with ~83,972 monthly package downloads; last push 0d ago; 100 releases/12mo; 168 open issues. Manual review: fast release churn — pin versions; 'reduces hallucinated logic' is GitHub positioning, no independent benchmark. License: MIT.
★ 140,964 GitHub stars
How the trust score is built
Full methodology →- Adoption · stars/downloads, log-scaled vs peers
- 8.8/10
- Maintenance · recency, releases, issue load
- 10.0/10
- Openness · license + docs
- 10.0/10
- Manual review · sanity read + red-flag check
- 9.0/10
- · 140,964 GitHub stars (log-scaled vs catalog peers)
- · blended with ~83,972 monthly package downloads
- · last push 0d ago; 100 releases/12mo; 168 open issues
- · manual review: fast release churn — pin versions; 'reduces hallucinated logic' is GitHub positioning, no independent benchmark
Sources
Every fact above was checked against these sources. Where a claim comes from the provider and couldn't be independently verified, the text says so.