SDD AI - Repository with documentation about specdriven.sh
SaferSkills independently audited specdriven (Agent Skill) and scored it 100/100 (green). The audit ran 55 deterministic rules across Security, Supply Chain, Maintenance, Transparency, and Community; it found 0 high-severity and 0 lower-severity findings. The full rule-by-rule trace and per-finding evidence are below. Free, methodology-open.
Findings & checks · 0 flagged
Every scanned point with the score it earned and what moved between them.
First recorded scan — no prior version to compare against.
The primary manifest — the file an agent reads to learn what this artifact does.
<div align="center"> <img src=".github/assets/header.svg" alt="specdriven.sh - Spec Driven Development" width="100%"/> </div>
<p align="center"> <a href="https://specdriven.sh"><img src="https://img.shields.io/badge/AI%20Agent%20Engineering-active-22d3ee?style=flat-square&labelColor=0a0a0a" alt="AI Agent Engineering"/></a> <a href="https://github.com/tomaszczechowski/specdriven-specs/tree/main/content/specs"><img src="https://img.shields.io/github/directory-file-count/tomaszczechowski/specdriven-specs/content%2Fspecs?type=dir&label=specs&style=flat-square&color=22d3ee&labelColor=0a0a0a" alt="Specs in catalog"/></a> <a href="https://github.com/tomaszczechowski/specdriven/commits/main"><img src="https://img.shields.io/github/last-commit/tomaszczechowski/specdriven?style=flat-square&labelColor=0a0a0a" alt="Last update"/></a> <a href="https://www.npmjs.com/package/specdriven"><img src="https://img.shields.io/npm/v/specdriven?style=flat-square&labelColor=0a0a0a&color=22d3ee" alt="npm version"/></a> <a href="./LICENSE"><img src="https://img.shields.io/github/license/tomaszczechowski/specdriven?style=flat-square&labelColor=0a0a0a" alt="License"/></a> </p>
<p align="center"> <a href="https://github.com/tomaszczechowski/specdriven-specs"><img src="https://img.shields.io/badge/Browse_Specs-22d3ee?style=for-the-badge&logo=github&logoColor=white&labelColor=0a0a0a" alt="Browse Specs"/></a> <a href="https://specdriven.sh"><img src="https://img.shields.io/badge/%F0%9F%8C%90_Live_Catalog-f5f5f5?style=for-the-badge&logoColor=0a0a0a&labelColor=0a0a0a" alt="Live Catalog"/></a> </p>
<p align="center"> <a href="https://github.com/sponsors/tomaszczechowski"><img src="https://img.shields.io/badge/Sponsor-on%20GitHub-ea4aaa?style=for-the-badge&logo=githubsponsors&logoColor=white" alt="Sponsor"/></a> <a href="https://www.buymeacoffee.com/tomaszczechowski"><img src="https://img.shields.io/badge/Buy%20Me%20a%20Coffee-fbbf24?style=for-the-badge&logo=buymeacoffee&logoColor=0a0a0a" alt="Buy Me a Coffee"/></a> </p>
Production project blueprints with AI built in. Skip the three days of stack debates. Install an opinionated tech stack, architecture, file layout, and references to the AI workflows that fit — all from a single CLI command.
<p align="center"> <img src="demo.gif" alt="Taskforge CLI demo" width="80%" /> </p>
# Install a complete project spec
npx specdriven add nextjs-saas
# Or pick interactively
npx specdriven add
# Browse the catalog from the terminal
npx specdriven list
npx specdriven find nextThat's it. The CLI fetches the spec from the public catalog and installs it into your agent's config directory — Claude Code, Cursor, Copilot, Codex, and 50+ others.
Spec Driven Development (SDD) is a methodology where AI-generated specifications drive your project from day one. Rather than bolting AI assistance onto an existing workflow, SDD treats specs as the source of truth — they describe your stack, your architecture, and the AI skills that come bundled with them.
A spec is the production-grade scaffold for a real project: opinionated tech, real architecture, file structure, and the AI workflows that pair with it. Each spec ships:
skills/<name>/, installed alongside the specnpx skills add ... commands the user can copy-paste through the open skills ecosystem CLIInternal skills are spec-specific (e.g. a code-reviewer tuned for the spec's stack). External skills point to generic skills already maintained elsewhere — specdriven never duplicates that content.
This repo is the docs hub for the project. The actual code and specs live in dedicated repos:
| Repo | Purpose |
|---|---|
| [specdriven-specs](https://github.com/tomaszczechowski/specdriven-specs) | Community catalog of project specs (with bundled skills) |
| [specdriven.sh](https://specdriven.sh) | Web catalog, browse UI, and online docs |
The CLI source lives in a private repo and is published to npm as specdriven.
Install once and use anywhere:
npm install -g specdriven
# or run ad-hoc via npx specdriven <command>Global helpers:
specdriven --version # print the installed version
specdriven --help # top-level help
specdriven <command> --help # per-command help (e.g. specdriven add --help)add — install a specspecdriven add [slug] [options]specdriven add nextjs-saas # install a spec
specdriven add # interactive prompt
specdriven add nextjs-saas --global # install into user-level config
specdriven add nextjs-saas --agent cursor # target a specific agentAfter a successful install, the CLI prints any external skills paired with the spec as a list of npx skills add ... commands you can copy-paste.
| Option | Description |
|---|---|
-a, --agent <id> | Target agent (e.g. claude-code, cursor) |
-g, --global | Install into the user-level config dir |
-d, --dest <path> | Override the destination root |
-f, --force | Overwrite if the destination already exists |
-y, --yes | Skip prompts; require all args/flags |
find — search the catalogPartial-substring search across slug, title, description, and tags. Omit the query to list everything.
specdriven find [query]specdriven find next # matches "nextjs-saas", "nestjs-api", etc.
specdriven find # list every spec in the catalog_No options — this is a read-only search command._
list (alias: ls) — browse the catalogLists the catalog alphabetically, 20 entries per page.
specdriven list
specdriven ls # alias form_No options — output is paginated automatically._
init — scaffold a new specCreates a new spec from the official template, ready to author and contribute back upstream.
specdriven init [slug] [options]specdriven init my-new-spec
specdriven init my-stack --global
specdriven init # interactive
specdriven init review-bot --agent cursor # scaffold targeting cursor's layoutIf slug is omitted, it defaults to spec-example.
| Option | Description |
|---|---|
-a, --agent <id> | Target agent (claude-code, cursor, etc.) |
-g, --global | Scaffold into the user-level config dir |
-d, --dest <path> | Override the destination root |
-f, --force | Overwrite if the destination already exists |
-y, --yes | Skip prompts; require all args/flags |
remove (alias: rm) — uninstallRemoves a previously installed spec from your agent's config directory. Asks for confirmation unless --yes is passed.
specdriven remove <slug> [options]specdriven remove nextjs-saas
specdriven rm nextjs-saas --yes # alias + skip confirmation
specdriven remove my-stack --global # remove from user-level config
specdriven rm nextjs-saas --agent cursor # target a specific agent's path| Option | Description |
|---|---|
-a, --agent <id> | Target agent |
-g, --global | Remove from user-level config |
-d, --dest <path> | Override the destination root |
-y, --yes | Skip confirmation |
specdriven works with 55 AI coding agents out of the box. Pass --agent <id> to target a specific one, or let the CLI auto-detect from your project. Agent IDs align with the open skills ecosystem so installed content interops across tools.
| Agent | ID |
|---|---|
| Claude Code | claude-code |
| Cursor | cursor |
| GitHub Copilot | github-copilot |
| Codex | codex |
| OpenClaw | openclaw |
<details> <summary><b>Show all 55 supported agents</b></summary>
<br/>
| Agent | ID |
|---|---|
| AdaL | adal |
| AiderDesk | aider-desk |
| Amp | amp |
| Antigravity | antigravity |
| Augment | augment |
| Cline | cline |
| Code Studio | codestudio |
| CodeArts Agent | codearts-agent |
| CodeBuddy | codebuddy |
| Codemaker | codemaker |
| Codex | codex |
| Command Code | command-code |
| Continue | continue |
| Cortex Code | cortex |
| Crush | crush |
| Cursor | cursor |
| Claude Code | claude-code |
| Deep Agents | deepagents |
| Devin for Terminal | devin |
| Dexto | dexto |
| Droid | droid |
| Firebender | firebender |
| ForgeCode | forgecode |
| Gemini CLI | gemini-cli |
| GitHub Copilot | github-copilot |
| Goose | goose |
| Hermes Agent | hermes-agent |
| IBM Bob | bob |
| iFlow CLI | iflow-cli |
| Junie | junie |
| Kilo Code | kilo |
| Kimi Code CLI | kimi-cli |
| Kiro CLI | kiro-cli |
| Kode | kode |
| MCPJam | mcpjam |
| Mistral Vibe | mistral-vibe |
| Mux | mux |
| Neovate | neovate |
| OpenClaw | openclaw |
| OpenCode | opencode |
| OpenHands | openhands |
| Pi | pi |
| Pochi | pochi |
| Qoder | qoder |
| Qwen Code | qwen-code |
| Replit | replit |
| Roo Code | roo |
| Rovo Dev | rovodev |
| Tabnine CLI | tabnine-cli |
| Trae | trae |
| Trae CN | trae-cn |
| Universal | universal |
| Warp | warp |
| Windsurf | windsurf |
| Zencoder | zencoder |
Missing your agent? Open an issue and we'll add it.
</details>
This repo houses docs and the project hub — it doesn't accept spec submissions directly. Open contributions go to the catalog:
CI validates frontmatter and content, a maintainer reviews, and on merge your contribution syncs to specdriven.sh within an hour.
For doc improvements, typo fixes, or issues with the CLI itself, open a PR or issue on this repo.
If specdriven saves you time, consider supporting development:
<p align="left"> <a href="https://github.com/sponsors/tomaszczechowski"><img src="https://img.shields.io/badge/Sponsor-on%20GitHub-ea4aaa?style=for-the-badge&logo=githubsponsors&logoColor=white" alt="Sponsor"/></a> <a href="https://www.buymeacoffee.com/tomaszczechowski"><img src="https://img.shields.io/badge/Buy%20Me%20a%20Coffee-fbbf24?style=for-the-badge&logo=buymeacoffee&logoColor=0a0a0a" alt="Buy Me a Coffee"/></a> </p>
v2.x — actively developed. v2.0.0 introduced the spec-only model (skills are now bundled inside specs or referenced as external commands). See the CHANGELOG for full migration notes.
MIT — Copyright (c) 2026 Tomasz Czechowski.
~30 seconds. Free. No account. Every finding cites a rule and a line of evidence.