<div align="center">

AI Dev Kit: Agent Skills, Coding Standards and Claude Code Mods

Agent skills, optional MCP tools, and layered coding standards for AI-assisted development — Claude Code, OpenCode, Codex, and any coding agent that reads markdown.

AI Dev Kit is a free, MIT-licensed toolkit that makes AI coding agents behave consistently and spend fewer tokens. It ships 64 reusable SKILL.md agent skills (53 for Claude Code, 11 for OpenCode models), layered coding standards for CSS, HTML, JavaScript, TypeScript, accessibility, SEO, performance, Git and QA, 18 opt-in slash commands, 11 Claude Code mods that enforce the rules in code, and opt-in local MCP servers. A full Claude Code install adds about 3,900 tokens of always-on context (estimated).

Portable SKILL.md agent skills, framework-agnostic coding standards, opt-in slash commands, git hooks and Claude Code mods that enforce the rules in code, opt-in free/local MCP servers (UI prototyping, image reading, image generation, file pre-filtering), an opt-in offline text-to-speech add-on that reads Claude's replies aloud, and opt-in reply personas, for teams building with AI pair programmers. One clone. Drop two folders into any project. Every session works your way — and gets better at it with use.

---

Listed under AI Hub → Skills on Three.js Resources — a curated directory for Three.js AI tools.

</div>

---

Why use it

• Consistent sessions. The standards chain loads for each file you edit (the optional standards-chain mod makes that mechanical), and skills load when the agent recognises their trigger, so session ten behaves like session one. Skill loading is model-judged, not guaranteed; the mods are what enforce the critical gates.
• Fewer wasted tokens. Skills load only when relevant, reads are targeted, cheap models take the cheap work, and a script finds violations so the agent does not have to.
• Standards that are enforced, not just written down. Rules are testable, the agent self-checks before a commit, and gaps are flagged instead of improvised.
• Cleanup of old or shared code. code-audit counts violations across a whole project and lists the exact lines, so the agent reads only what is flagged.

Index

• Install into a project — clone it, run it, answer a few questions
• Everything the kit does, at a glance — what happens by itself vs. what you ask for
• What's in the kit — skills/ and coding-standards/, the two folders the kit is built from
• Complete skills catalog — every skill, by category, with its purpose
• Slash commands — every command you type, and when to reach for it
• Optional AI tools: free and local MCP servers — UI prototyping, image reading/generation, local file prefiltering
• Voice, narration and personas — opt-in text-to-speech for replies, and reply personas
• Claude Code mods (all ten) — the plug-ins that enforce rules in code instead of relying on a skill loading
• Scripts and git hooks — what every script does, and setting up git hooks for human commits
• Maintaining the kit — keeping this repo and its installs in sync over time
• Where things live, and how to turn things off — every file the kit writes, and how to remove each one
• Key ideas — the design philosophy behind the kit
• Repository layout — the folder tree, annotated
• Recent additions — latest changes, newest first
• Frequently asked questions
• Contributing and support
• License

Install into a project

Fastest path, if you're already working inside an AI coding chat (Claude Code, OpenCode, Codex, or any agent that reads markdown):

git clone https://github.com/AftabIbrahimKazi/ai-dev-kit.git
cp -r ai-dev-kit/skills ai-dev-kit/coding-standards your-project/

On Windows PowerShell: Copy-Item -Recurse ai-dev-kit/skills, ai-dev-kit/coding-standards your-project/

Then, in your-project, say:

read skills/README.md and install

The core installer copies markdown files only: no packages, no build step. Opt-in add-ons that need a tool (for example narrate, which uses Piper for speech) install it only when you ask.

Working from a terminal instead, with no AI driving it (or in OpenCode/Codex, which have no click-to-answer chat widget)? The same installer runs as a plain script with its own numbered-menu prompts — see the complete walkthrough below.

1. Copy skills/ — and coding-standards/ if the project should carry the standards — into the project root. (Or keep one clone of this repo anywhere and point every project at it with --library <path>; either way, it's the same install-kit.cli.js doing the work.)

2. Open your AI coding tool in that project and say:

   read skills/README.md and install

3. Choose everything, pick, auto-detect (scans the project and cites the signal before pinning anything), or a stack preset (name the stack directly — Astro, Next.js, Laravel — no scan needed). The installer detects which coding tool you're in, copies the chosen skills to that tool's skill root (.claude/skills/ for Claude Code, .opencode/skills/ for OpenCode), wires the standards into that tool's session file (CLAUDE.md or AGENTS.md respectively), mechanically re-verifies the result, and reports what was installed — or exactly what still needs fixing before it's reported done. install-kit itself is never copied into your project — it always runs from wherever the library lives, against --root <target>, so there's nothing left behind to clean up afterward.

How you answer those questions depends on where you're running it — pick based on that, not preference:

| Where | How you answer | Best for | |---|---|---| | Claude Code chat (recommended default) | Click multiple-choice options at every decision point — mode, tool confirmation, which skills, comment style, context code. Typing only comes up for a hand-picked skill list or an override's specifics. | Anyone already working in Claude Code — fewest keystrokes, nothing to remember. | | OpenCode, Codex, or a plain terminal | The installer asks each question as a plain numbered list — type the number of your answer and press Enter. No extra software to install first. | Tools with no click-to-answer chat widget, or installing with no AI in the loop at all — see the walkthrough just below. | | Direct CLI flags (--tool, --skills, --standards, ...) | No questions at all; fully scripted. | CI, re-installs across many projects, or anyone who already knows exactly what they want. |

None of these change what gets installed — only how you're asked. If you're not sure, let whichever tool you're already in drive it; it'll default to the right mode for where it's running.

Installing from a terminal (PowerShell, Command Prompt, Mac/Linux shell) — a complete walkthrough

If you've never used a terminal before, this is every single thing to type, in order. It's the same on Windows, Mac, or Linux except step 1.

1. Open a terminal. Windows: click Start, type PowerShell, press Enter. Mac: open Terminal from Spotlight. Linux: open your usual terminal app.

2. Check that Node.js is installed. Type this and press Enter:

   node --version
   

   If you see something like v22.11.0, you're set — skip to step 3. If you see an error instead, Node.js isn't installed: go to nodejs.org, download the button labeled LTS, run the installer with its defaults, then close and reopen your terminal and try step 2 again.

3. Download a copy of this kit. Type these three lines one at a time, pressing Enter after each:

   cd Documents
   git clone https://github.com/AftabIbrahimKazi/ai-dev-kit
   cd ai-dev-kit
   

   (If you don't have git, download the ZIP from the green "Code" button on this repo's GitHub page, unzip it, and cd into the unzipped folder instead.)

4. Run the installer against your actual project. Which command to type depends on where the ai-dev-kit folder ended up.

   If the ai-dev-kit folder is sitting inside your project folder — for example your-project\ai-dev-kit, because you opened a terminal inside your project and ran git clone right there instead of following step 3 above — that's actually the single most common setup, and you don't need to point at a project path at all. While your terminal is still inside the ai-dev-kit folder, just run:

   node skills/meta/install-kit.cli.js install
   

   The installer notices it's running from inside its own folder, uses the folder one level up (your project) automatically, and prints what it assumed — so it's never a silent guess. It only stops to ask for a path if there's no safe folder to assume. If it ever guesses wrong, add --root "C:\full\path\to\your-project" to say so explicitly.

   Otherwise — the ai-dev-kit folder sits somewhere else, separate from your project (like Documents, if you followed step 3 above exactly) — you have to tell the installer which project folder to use with --root. There are two ways to type that command. They do the exact same thing — which one works depends only on whether you closed your terminal since step 3.

   Option A — you never closed the terminal window from step 3. You're still sitting inside the ai-dev-kit folder you just downloaded, so the installer script is right where your terminal is already looking. Type:

   node skills/meta/install-kit.cli.js install --root "C:\full\path\to\your-project"
   

   Replace C:\full\path\to\your-project with the folder of the project you want the kit installed into. Don't type that path by hand — open the folder in File Explorer (or Finder/your file manager) and drag it into the terminal window; the full path gets pasted in for you automatically.

   Option B — any other time (you closed that terminal and opened a new one, restarted your computer, or opened a terminal inside your project folder instead of the kit folder). The command from Option A will fail here with an error like Cannot find module '...\skills\meta\install-kit.cli.js' — that happens because it only looks for the installer relative to whatever folder your terminal happens to be sitting in right now, and that folder is no longer ai-dev-kit. Instead, give the command the full path to both folders, so it doesn't matter where your terminal is sitting:

   node "C:\full\path\to\ai-dev-kit\skills\meta\install-kit.cli.js" install --root "C:\full\path\to\your-project"
   

   Same trick for both paths: drag the ai-dev-kit folder into the terminal window to paste its path whe