Rank #07 · Daily · DevTools
spec-kit: GitHub's Toolkit for Spec-Driven Development with Coding Agents
Install Specify, pick a process (build / fix / assess), and run agent skills that turn a written spec into a plan, tasks, implementation, and a convergence check.
What it does
Spec Kit is GitHub's open toolkit for giving AI coding agents structured processes, reusable templates, and documented outcomes. It is not a single slash-command dump: SDD ships in core; bug fixing and idea assessment are bundled extensions; presets, workflows, catalogs, and bundles let teams customize or bring their own process.
Entry points from the README:
| Need | Process | Outcome | |---|---|---| | Build a feature/app | Spec-Driven Development | Spec → plan → tasks → implement → converge | | Repair broken behavior | Bug fixing (specify extension add bug) | Assessed cause, scoped fix, verified/partial/failed verdict | | Decide if an idea is worth building | Idea assessment (specify extension add assess) | Evidence-backed go / needs-clarification / kill | | Something else | Extensions + presets + workflows | Your process, assembled from primitives |
Minimal SDD loop (agent chat skills; Copilot syntax shown in the README):
/speckit-constitution ...
/speckit-specify ...
/speckit-plan ...
/speckit-tasks
/speckit-implement
/speckit-convergeInstall path: Python 3.11+, uv tool install specify-cli, then specify init my-project --integration <agent>. Integrations cover multiple coding agents (docs list keys and invocation syntax). Organizations can host private catalogs, work air-gapped with local sources, and enforce standards via presets. Docs live at github.github.com/spec-kit.
Who it's for
- Teams adopting spec-driven development with Copilot, Claude Code, or other supported agents who want a shared constitution and repeatable command surface.
- Engineering orgs that need governed customization (presets, private catalogs, regulatory traceability) rather than ad-hoc prompt packs.
- Individuals who want a documented bug-fix or idea-assessment workflow without inventing file layouts from scratch.
Less useful if you only want a one-shot prompt and no process artifacts under .specify/.
Stars / language / license
147,321 stars · Python · MIT
Data verified: 2026-11-05 09:31 Asia/Shanghai (GitHub API get_repository + README get_file_contents)
Trending note: +808 stars today (Trending selection captured 09:26 Asia/Shanghai)
Category tags
DevTools · spec-driven-development · coding-agents · GitHub · Python · CLI · Copilot
How it differs from similar projects
Compared with raw AGENTS.md / skills dumps. Spec Kit is a process toolkit with CLI install, versioned commands, extensions, and convergence artifacts — not a folder of Markdown prompts alone.
Compared with PRD generators or ticket bots. Those often stop at a document. Spec Kit carries the spec through plan → tasks → implement → converge (and parallel bug/assess tracks) inside the agent chat loop.
Compared with enterprise SDLC platforms. Spec Kit stays local/agent-centric and MIT-licensed; enterprise depth comes from presets, bundles, and private catalogs you host — still agent-driven, not a mandatory SaaS ALM.
TidyHow has not measured whether SDD reduces defect rates versus freeform coding agents. Treat the value as process structure and shared vocabulary, not a published productivity benchmark.