[rotundtapir]

Fixes a wasm-only runtime crash that 500's web e2e caught on its adoption PR (500#36) and that no JVM test can see.

With sealed hierarchies, encodeToString<ClientMessage>(…)'s reified lookup found the generated serializer on every target. Now that the bases are plain interfaces, the JVM still resolves one reflectively at runtime — but Kotlin/Wasm has no such fallback and throws Serializer for class 'ClientMessage' is not found … explicitly declared serializer should be used for interfaces on the client's very first outbound frame. Every JVM suite and android-e2e stayed green while the browser build died on connect.

ClientMessageSerializer / ServerMessageSerializer are now the API, and the client and the server transport both use them. There are no reified-interface call sites left in either module, which is what stops this recurring.

The goldens are unchanged and still pass, which is the point worth taking away: open polymorphism is byte-identical on the wire but not lookup-identical across targets. Recorded in CLAUDE.md, along with the fact that this repo has no wasm unit test (browser tests need a real Chrome), so the games' Playwright suites are the only stage that exercises the wasm client — which is why they gate merges.

Verified: :cardkit-net:jvmTest, :cardkit-server:test, :cardkit-net:compileKotlinWasmJs. The real proof is 500#36 and euchre going green on their web e2e after re-pinning; I'm running 500's locally now.

🤖 Generated with Claude Code

https://claude.ai/code/session_019EH6XkA2oEhv4YnMXt8hbJ

[rotundtapir]

Reviewed: precisely the fix from the 500#36 diagnosis — ClientMessageSerializer/ServerMessageSerializer as published PolymorphicSerializers, frame call sites switched off the reified form, and the KDoc carries the cross-target rationale so the rule survives its discoverer. Single commit authored/pushed by rotundtapir, CI green. Merging; 500#36 re-pins and its web e2e becomes the live proof on the wasm target.