• TODO: ktichen sink counter latency #5051
• fix(rivetkit): use engine actor stop threshold for shutdown #5049
• DO NOT MERGE: serverless restart race condition #4997
• fix(rivetkit): exit pid1 after signal shutdown #5045
• fix(rivetkit-core): decrement active actor metrics #5044
• fix(envoy-client): dont abort drain after 20s #5043
• temp: patch wasm-pack to point to latest binary #5037
• fix(rivetkit): require engine ping for health #5022
• feat(rivetkit): add current actor metrics #5020
• feat(rivetkit): expose low-cardinality metrics routes #5019
• docs(serverless): document timeout tuning #5018
• feat(kitchen-sink): add load testing harness #5017
• fix(rivetkit): expose client actor metadata #5016 👈 (View in Graphite)
• fix(runtime): avoid bounded dispatch channel stalls #5015
• fix(sqlite): fence remote requests by generation #5014
• feat(rivetkit-core): expose metrics endpoint #5013
• WIP: counter latency monitor #5011
• main

This stack of pull requests is managed by Graphite. Learn more about stacking.

Code Review: fix(rivetkit): expose client actor metadata

Overview

This PR does three things:

1. Adds the actor metadata field to the AsyncAPI spec (catching the docs up to the existing implementation)
2. Adds a CLAUDE.md note on client/server protocol versioning assumption
3. Pre-populates #actorId in the ActorConnRaw constructor for getForId resolution states so the ID is available before the WebSocket Init message arrives

Small, focused change with good intent. A few items worth addressing before marking ready.

Issues

1. Test does not exercise the constructor fix (medium)

The new test uses client.counter.getOrCreate(...), which produces a getOrCreateForKey resolution state, not a getForId state. The constructor fix (if ("getForId" in actorResolutionState)) is therefore never exercised. actorId in that test comes from the WebSocket Init message (line 728), which already worked before this PR.

To cover the new path, add a case that:

1. Resolves an actor ID first via handle.resolve()
2. Opens a getForId connection via client.getForId(actorId)
3. Checks connection.actorId before await connection.ready - that is the behavior the constructor change unlocks

2. actorId JSDoc still says "for testing purposes" / @internal (low)

actor-conn.ts:1463-1469: The getter is marked @internal with "for testing purposes". If this is being promoted to a stable public API for users to read the resolved actor ID, the JSDoc needs updating. If it stays @internal, callers outside tests should not rely on it.

3. asyncapi.json generation field type is redundant (low)

Both schema additions declare anyOf: number or integer/int64. A JSON Schema integer is a strict subset of number, so the anyOf always reduces to number. The intent seems to be expressing the TypeScript z.union(z.number(), z.bigint()) distinction, but that is a JS runtime type difference, not a JSON Schema distinction. Using just integer/int64 alone would be cleaner and avoid confusing generator tooling.

4. CLAUDE.md note placement is slightly off (low)

The new bullet is placed in the BARE/vbare schema section, but the statement ("client sends its latest request protocol version; servers handle older-client compatibility") is a general protocol versioning principle, not BARE-specific. Consider moving it to a higher-level section.

What looks good

• The core fix is correct: for a getForId resolution state the actor ID is known at construction time, so populating #actorId early is the right call.
• Existing reconnect/clear logic (#clearResolvedActorIdentity + Init message re-set) remains intact and consistent with the early initialization.
• The actor schema additions in asyncapi.json match ActorSpecifierSchema in client-protocol-zod.ts exactly (actorId required, generation required, key optional).
• The HttpResponseErrorSchema parity (second asyncapi.json hunk) is also correct.