Open‑source agent‑readiness tester

A shared, MIT‑licensed platform (Scope) runs repeatable experiments that trace how AI coding agents discover and use a product (SDKs, CLIs, plugins), surfacing documentation gaps, tooling bugs, and behavioural changes after updates. Teams can self‑host the system to evaluate private codepaths, collect traces, and turn agent failures into concrete engineering tasks. — If adopted broadly, the tool can set de facto standards for 'agent‑readiness', reshape product design priorities toward agent users, and centralize who controls interoperability and evaluation norms.