Safe AI Coding Rules
Keep your AI coding assistant on-track and stop it from doing risky things without asking.
Your first result: your assistant follows your project rules and asks before risky actions.
$29
One-time · instant download
Coming soon. Leave your email and we'll tell you the day it opens.

What you get
What's inside the kit
Ready-to-use AI rules for your coding assistant
Tell your assistant your project's conventions, and to plan and ask before it acts.
Add-on packs for frontend, API, database, infrastructure and PRs
Install only the rules that fit your project.
Four built-in skills
Security review, writing tests, safe refactoring and a ship checklist, each on request.
Two read-only helper assistants
A code reviewer and a test runner that report back without changing your code.
A safety hook that blocks risky commands
Helps stop things like force-pushes, reading secret files or production deploys before they run.
A one-command installer
Shows you what it will add first, and never overwrites files you already have.
Example projects (Node.js and Python)
See everything set up and working before you add it to your own project.
How to put it to work
How to get started
Follow these steps in order. After you buy, the same steps appear as a checklist in your kit viewer, with links to the exact guide for each step.
- 1
Commit your project
2 minSave your work in Git first so you can review exactly what the installer adds.
- 2
Preview the install
2 minRun the installer for your stack (Node or Python). It lists every file it would add; nothing is written yet.
- 3
Install it
1 minRun the same command with --apply. Files you already have are skipped and left untouched.
- 4
Fill in your project facts
5 minIn the new CLAUDE.md / AGENTS.md, check your stack and your install, test and lint commands.
- 5
Check it's working
3 minIn Cursor, open Settings → Rules. In Claude Code, run /agents, or ask it to force-push and watch the hook block it.
- 6
Use it day to day
ongoingAsk for "security-review on my diff", "use the code-reviewer subagent" or "ship checklist". These report only and change nothing.
Is it right for me?
Who it's for
A great fit if…
- Solo developers and indie hackers using Cursor or Claude Code
- JavaScript/TypeScript or Python projects
Probably not for you if…
- You don't use an AI coding assistant
- You want a sandbox: the safety hook is a safety net, not a sandbox
Questions
Frequently asked questions
Will it overwrite my existing rules?
No. The installer only previews unless you add --apply, never overwrites existing files and never runs git. Merge sections from the templates yourself if you want them.
Do I need any accounts or account keys?
No. It needs Node.js 18.17+, Git and Cursor or Claude Code.
Can it still make mistakes?
Yes. AI assistants can still get things wrong, and the hook is pattern-based. You review and approve everything.
How do I remove it?
Delete the added .claude/ and .cursor/rules files, CLAUDE.md and AGENTS.md, or revert them with Git.
Good to know before you buy
- Provided as is, without warranty of any kind; liability limited as set out in the license.
- No promise of any particular outcome, time savings or income.
- AI agents can still make mistakes: you are responsible for reviewing and approving all agent output, commands and changes, and for keeping backups.
- The guard hook is pattern-based: a safety net, not a sandbox.
Technical details (for the technically curious)
Technical words on this page, explained
- AI agent:
- An AI assistant that can take steps for you (like running a command or calling a tool), not just chat.
- Repo:
- A project folder tracked with Git, so every change is saved and can be undone.
- Hook:
- A small script that runs automatically at a set moment, for example to block a risky command.
- CLAUDE.md / Cursor rules:
- Plain text files of instructions your AI coding assistant reads before it starts working.
- CI / GitHub workflow:
- Automatic checks that run every time code changes, so problems are caught early.
- Env file / secrets:
- Private settings like passwords and keys. They must never be shared or saved in public code.
Formerly / technical name: Agent Rules + Skills Pro Kit
Full description
A tested set of project rules, skills, hooks and subagents. They help keep AI coding agents on your conventions, reduce drift into unrelated changes, and help block common risky actions such as force-pushes, reading .env files or publishing packages. Approval-first: agents plan, ask, and only then act. You review all agent output.
Full parts list
- CLAUDE.md / AGENTS.md templates
- Cursor rules: core, security, git + Node.js and Python stack rules
- Claude Code skills: security-review, write-tests, safe-refactor, ship-checklist
- Read-only subagents: code-reviewer, test-runner
- PreToolUse safety hook
- Installer (dry-run by default, never overwrites)
- Example repo per stack with CI
- Updates as released (no fixed schedule)
Requirements
- Node.js 18.17+
- Cursor or Claude Code
- Git
Tested on
- Node 20.19.2 / npm 11.21.0
- Python 3.13.5
- Linux 6.12.94+
Tested locally with static format checks. Not yet run inside Cursor or Claude Code; example CI not run on GitHub. (2026-10-28)
What's in the download (88 files)
- 📁 examples/ (41 files)
- 📁 scripts/ (1 file)
- 📁 templates/ (31 files)
- 📁 tests/ (8 files)
- 📄 .gitignore
- 📄 CHANGELOG.md
- 📄 LICENSE
- 📄 package.json
- 📄 README.md
- 📄 REVIEW_STATUS.md
- 📄 SETUP.md
Version and changes
v0.2.0 · tested 2026-10-28
- 0.2.0 · 2026-10-30: Curated packs (5) and new infra/data guard rules; Node 24 test fix.
- 0.1.0 · 2026-10-28: Initial release: templates, Cursor rules, skills, subagents, guard hook, installer, Node and Python examples with CI.