TaiwildLab is an independent lab. We build diagnostic kits, audit scripts and automation tools — born from real problems, debugged in real use.
Tivoli is the lab where we navigate the chaos of AI agent tooling. We build diagnostic kits from real engineering problems — debugged in real use, published openly.
The AI agent ecosystem is overflowing with tools, layers and "magical" solutions. Tivoli diagnoses what actually breaks in production — and ships the smallest fix that works.
Every decision, mistake and bug ends up in the Substack log. Real time, no editorial filter — including the dead ends.
Every kit is one Python script. No virtualenvs, no API keys, no SaaS lock-in. Works the same on your laptop as in ours.
Real-time system journal. What works, what fails, and why.
Built in the lab from real problems. Free with email.
Direct feedback from you. Two minutes of yours is worth more than any metric.
What's missing? What would you improve? Is there a tool you need? We read everything.