Try it yourself in ~10 minutes

Clone the repo, run npm run demo, and investigate a fictional checkout deploy question in Cursor or Copilot. No live call required.

01

Clone the repo

Open source under GPL-3.0. No account or API key required.

git clone https://github.com/iman-suherman/ai-engineering-workbench.git
cd ai-engineering-workbench
02

Run the demo script

Scaffolds a sample investigation with a fictional checkout enquiry.

npm run demo
03

Open in Cursor

Open the repo root. The agent reads AGENTS.md and the service registry automatically.

cursor .
04

Paste the demo prompt

Copy examples/demo-cursor-prompt.md into Cursor Chat (or use prompts/investigate.md).

cat examples/demo-cursor-prompt.md
05

Review the output

Finding, Evidence, Risk, Recommendation — you decide what to send. Nothing goes out automatically.

What you are testing

  • Service registry navigation for multi-repo enquiries
  • Evidence-first investigation prompts (Finding / Evidence / Risk)
  • Human-in-the-loop — you review before anything is sent

Prefer a guided session with your real repos? Book a live demo — ~30–45 min with Iman.