The audit observed src/proof.ts taking a snapshot and then reading SceneGraph, active-app and media state again for the summary. This is a hypothesis about duplicate acquisition, not proof that the reads are unnecessary.
Inspect current source and establish which observations intentionally require later/fresher state or different device-info detail. Measure request count and elapsed time with deterministic transport and, when authorized, a representative Roku. Record the exact installed package and device workload.
Only reuse an observation if freshness and public summary/artifact semantics remain correct. Verify installed API/CLI consumers and affected proof output after any change. Do not parallelize device calls or add caching solely because adjacent reads look similar. A documented no-change result is valid.
The audit observed src/proof.ts taking a snapshot and then reading SceneGraph, active-app and media state again for the summary. This is a hypothesis about duplicate acquisition, not proof that the reads are unnecessary.
Inspect current source and establish which observations intentionally require later/fresher state or different device-info detail. Measure request count and elapsed time with deterministic transport and, when authorized, a representative Roku. Record the exact installed package and device workload.
Only reuse an observation if freshness and public summary/artifact semantics remain correct. Verify installed API/CLI consumers and affected proof output after any change. Do not parallelize device calls or add caching solely because adjacent reads look similar. A documented no-change result is valid.