glossary · term
agent-readiness test
A test of platform maturity. If an AI coding agent joined your company tomorrow, could it figure out how to deploy an application correctly without asking someone on the platform team? A failure says more about the platform than about the AI, because the same gaps already tax developers.
the test
Could an agent do each of these without asking anyone?
- 01Discover the approved path.
- 02Pick the right application pattern.
- 03Understand which controls apply.
- 04Interpret a failed pipeline.
- 05Find the authoritative documentation instead of a three-year-old example.
- 06Tell what is technically possible from what the organization wants developers to do.
in practice
How I use the term
The bar is higher than producing plausible YAML. An agent knows only what the platform makes available to it, so it cannot work around ambiguity the way people do every day.
If the correct path depends on tribal knowledge, people struggle with it too. If the documentation contradicts the automation, people eventually notice too. Agents make those inconsistencies easier to see.
I suspect most of our platforms would fail it today. In that sense, agent readiness may become a useful new measure of platform maturity.
in the essays · 4
Where it shows up
- Your Platform Has a New User: The Agent
Where the test comes from: the platform now has two personas.
- Context Is a Platform Capability Now
The first thing a failing platform is missing: trusted context.
- The Golden Path Needs an Intent Interface
The interface an agent-ready platform needs.
- Your Platform Is Teaching Your Engineers
What defaults and errors teach engineers and agents.
see: golden path · agent · context engineering