todlando -> doyle, cc hertz. Workspace sweep result, plus an instrument artifact I nearly misread as a red.

SWEEP: `cargo check --workspace --all-targets` on feat/304-remote-friction-product at 5853cb14 — EXIT 0, Finished in 30.90s, ZERO errors and ZERO warnings across every crate including each lib's own test module. So the E0061 in webserve.rs was the only instance of that class on my branch; it has no siblings hiding in another crate.
Carrying hertz's caveat rather than dropping it now that it is inconvenient: this green is about MY branch. By his own argument it says nothing about the merged tree, and the cross-lane case is caught only by an --all-targets check on the MERGED tree at assembly. I am not quoting this as merge evidence.

THE ARTIFACT, because it is the kind of thing that turns into a false incident report: my background task reported "failed with exit code 1". The check did not fail. My wrapper ended with `grep -cE "^error" <log>`, and grep exits 1 when it counts ZERO matches — so the task's exit was the exit of a grep that found no errors, i.e. the SUCCESS case rendered as a failure. WS_CHECK_EXIT=0 was captured inside the wrapper and is the real answer.
The rule it re-teaches, and I had it written down already: read the verdict from the PRODUCER's exit, never from the last thing in the pipeline. A wrapper whose final command is a counter inverts your result exactly when there is nothing to report. I will put the producer's exit last, or capture it explicitly, in every wrapper from here.

LANE STATE at 5853cb14: dfbfd287 (distinct effect seq per advisory arm, five arms, prediction green filed) and 5853cb14 (the six-line stale-port repair, subsumed by hertz's module and yours to resolve in his favor at assembly). Uncommitted and deliberate: docs/adr/0058 Amendment 1 draft, pending the operator's #300 fork.
Waiting on: hertz's producer-2 window, then my one attempt at the distinct-seq cell; and your ruling on whether 5853cb14 stays or reverts for a clean assembly base. Nothing of mine executing.