The 4 Levers Agent Diagnostic
Use Context, Tools, Loop, and Governance to diagnose one real AI agent failure in 15 minutes.
What's inside
- A concise framework for diagnosing agent failures through Context, Tools, Loop, and Governance
- Guided questions for applying each lever to one real agent run
- Failure-mode checkboxes for common problems: missing context, bad tool surfaces, runaway loops, weak approval gates
- A final diagnosis section to identify the smallest harness change to try next
Built by Damian Galarza, a software engineer with 15+ years of production experience who designs and debugs real agent harnesses.
Frequently asked
- Is it really free?
- Yes. You'll get the worksheet by email and be subscribed to the newsletter. Unsubscribe anytime.
- What format is it in?
- A short printable PDF worksheet. It is designed for one real agent failure, not a broad audit.
- Who is this for?
- Senior developers, tech leads, founders, and AI builders who have seen agents ignore context, call the wrong tool, loop, stall, or claim success too early.
- I'm already subscribed. Should I still enter my email?
- Yes. Use the same email address and Buttondown will send the diagnostic without creating a duplicate subscription.
Get the 4 Levers Agent Diagnostic
I'll send the PDF to your inbox. You'll also get practical notes on AI engineering and agent architecture. No spam. Unsubscribe anytime.