Developer lane
Tested kits for building with AI agents
Agent rules and skills for Cursor and Claude Code, MCP server starters, eval and guardrail harnesses, and a free pre-launch security checklist.
Tested, not bulk
Each kit is checked against current tool versions, with a dated changelog and an honest note on what wasn't tested.
Approval-gated by default
Agents draft, you approve. Read-only and dry-run defaults; risky actions need a human step.
Documented end to end
Requirements, setup, least-privilege scopes and limits, written down.
Developer kits

Developer Starter Bundle
The three developer kits to build, connect and safety-test AI assistants, for less than buying them separately.

App Safety Checklist
Check your AI-built app for the most common security mistakes before you launch it.

Safe AI Coding Rules
Keep your AI coding assistant on-track and stop it from doing risky things without asking.

Connect Your Data Starter
Let AI assistants like Claude and Cursor safely use your own data or tools, starting from a working, tested template.

AI Safety Test Kit
Test your AI assistant before each release and add a safety layer that logs, flags and can switch it off.
New to these words? Here's what they mean
- AI agent:
- An AI assistant that can take steps for you (like running a command or calling a tool), not just chat.
- MCP:
- Model Context Protocol: a common plug that lets AI apps like Claude or Cursor connect to your data and tools.
- LLM / AI model:
- The AI "brain" behind tools like ChatGPT or Claude. You usually rent it from a provider with an account key.
- Evals:
- Repeatable tests for an AI assistant: the same questions every time, so you can see if it got better or worse.
- Guardrails:
- Safety checks around an AI assistant that flag or block risky requests and answers.
- Repo:
- A project folder tracked with Git, so every change is saved and can be undone.
- CLI / terminal:
- Typing short commands into a text window on your computer instead of clicking buttons.
- 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.
Not a developer? Start with the plain-English guides →