Spec Kit
GitHub's open-source toolkit for spec-driven development — the specify CLI and /speckit slash commands that walk any coding agent from constitution to implementation.
Spec Kit (GitHub, MIT, ~111k stars since its September 2025 launch) productized spec-driven development: specify init scaffolds a project, then /speckit.constitution, .specify, .plan, .tasks, and .implement walk your coding agent through the pipeline — each phase emitting a markdown artifact that feeds the next. Works with 30+ agents including Claude Code, Copilot, and Cursor.
Spec Kit is GitHub's answer to the chaos question of agentic coding — piecemeal prompting produces piecemeal software — and the answer landed: ~111k stars in nine months made it the standard expression of spec-driven development. It doesn't replace your agent; it gives your agent a process.
Highlights
- The five-phase pipeline —
/speckit.constitution(principles) →.specify(the what/why, deliberately no tech stack) →.plan(architecture and stack) →.tasks(actionable, checkable list) →.implement— each phase a reviewed markdown artifact feeding the next. - Quality gates —
/speckit.clarifyinterrogates vague specs before planning;/speckit.analyzecross-checks artifacts for drift;/speckit.checklistgenerates requirement checklists;.taskstoissuesturns tasks into GitHub issues. - Agent-agnostic — 30+ integrations (Claude Code, Copilot, Gemini CLI, Codex, Cursor, Windsurf…); commands install as the agent's native slash commands or as skills.
- Extensible — extensions add commands/workflows (Jira, code review), presets override templates for compliance formats or house terminology, with project-local overrides.
- Brownfield too — built for greenfield, parallel exploratory implementations, and iterative modernization of existing systems.
In an AI-assisted workflow
uv tool install specify-cli --from git+https://github.com/github/spec-kit.git@v0.10.2
specify init my-project --integration claude
# then, inside Claude Code:
# /speckit.specify → /speckit.plan → /speckit.tasks → /speckit.implementThe artifacts are the point: a spec.md and plan.md you review before code exists, and that re-anchor any future session — the persistence that prompts lack.
NOTE
It moves fast: releases ship weekly-ish (v0.10.x as of June 2026), templates change between them, and the old --ai flag is gone — pin a tag, and prefer specify self upgrade for updates. Community extensions are third-party; review before installing.
Good to know
MIT, Python 3.11+ with uv, launched September 2, 2025 via the GitHub blog and snowballed as the antidote-to-vibe-coding narrative took hold. The methodology it implements — when specs pay, when they're theater — is our Spec-Driven Development guide; the lighter-weight in-Claude-Code version of the same instinct is plan mode plus plan-feature.
Frequently asked questions
- What is Spec Kit?
- GitHub's open-source toolkit that turns spec-driven development into slash commands for your existing coding agent. The specify CLI scaffolds the structure; then inside Claude Code, Copilot, Cursor, or 30+ other agents you run /speckit.specify (what and why), /speckit.plan (tech and architecture), /speckit.tasks (actionable list), and /speckit.implement — with a constitution file carrying project principles into every stage.
- How do I install Spec Kit?
- uv tool install specify-cli --from git+https://github.com/github/spec-kit.git@<release-tag> (Python 3.11+, uv, Git required), then specify init my-project --integration claude (or copilot, cursor…). Pin a release tag — templates evolve fast — and note older tutorials' --ai flag was removed in v0.10 in favor of --integration.
- Does Spec Kit write the code?
- No — your agent does. Spec Kit's contribution is structure: phase templates, the artifacts each phase emits, and gates like /speckit.clarify (de-vague the spec before planning), /speckit.analyze (consistency check between tasks and implementation), and /speckit.checklist ('unit tests for English'). It's the workflow layer; the model and agent stay yours.
Related
- Spec-Driven Development with AI AgentsWrite the spec, let the agent implement against it — the SDD workflow (spec → plan → tasks → implement), when it beats prompt-and-iterate, and the tooling.
- Vibe Coding in 2026: What It Is, When It Works, When It BitesAn honest guide to vibe coding — where prompt-and-accept development genuinely pays, where it accumulates risk, and the guardrails that make it professional.
- Plan FeatureExplore the codebase and produce an implementation plan for a feature.
- Breakdown TaskDecompose a task into an ordered checklist of small, verifiable steps.
- Claude CodeAnthropic’s official agentic coding tool that runs in the terminal, IDE, and web.
- Github CopilotGitHub’s AI pair programmer with inline completions and an agent mode.