Hi @RonTuretzky π β smithers maintainer here. We read the .smithers/ gateway embedding, the src/seam client, and the build backend closely β it's a solid integration, and it surfaced real upstream API gaps we're now fixing. This is the vibecode-room-side action list, smallest-effort-first. Happy to send PRs for any of it.
1 Β· Delete after bumping 0.23.0 β 0.31.x
You're pinned at 0.23.0 (package.json#L37, .smithers/package.json#L17). Two workarounds are already obsolete upstream:
While bumping: VIBERSYN_BUILD.md#L23-L25 still carries the stale global-install guidance β worth refreshing in the same PR.
2 Β· Staleness in the integration
- Stale spec β d1-fix-steer-loop.md#L14-L17 says the steer loop signals with a static
correlationId = "steer", but the current workflow already passes the dynamic ctx.input.correlationId (vibersyn-process.tsx#L167-L174). The spec describes a bug that no longer exists β worth updating before someone "fixes" it back.
- Private internals β the embedded self-issued RPCs (gateway.ts#L117-L145, smithers-client.ts#L177-L214) go through the private
routeRequest / internal connection shapes. They work today, but can shift on any minor β Β§3 has the sanctioned replacement in flight.
3 Β· Upstream gaps β fixes in flight on smithersai/smithers
Your embedding surfaced these. We're filing them from this review and will cross-link the numbers here as they land on the tracker.
4 Β· Proposed sub-issues
Compiled against vibecode-room@355e89d with every claim checked against smithers v0.31.0 source. We're happy to send PRs β the version bump is the natural first one.
Hi @RonTuretzky π β smithers maintainer here. We read the
.smithers/gateway embedding, thesrc/seamclient, and the build backend closely β it's a solid integration, and it surfaced real upstream API gaps we're now fixing. This is the vibecode-room-side action list, smallest-effort-first. Happy to send PRs for any of it.1 Β· Delete after bumping 0.23.0 β 0.31.x
You're pinned at 0.23.0 (package.json#L37, .smithers/package.json#L17). Two workarounds are already obsolete upstream:
rpcRawany-castSmithersGatewayClientexposes a publicrpcRawsince 0.31.0 β SmithersGatewayClient.ts#L319While bumping: VIBERSYN_BUILD.md#L23-L25 still carries the stale global-install guidance β worth refreshing in the same PR.
2 Β· Staleness in the integration
correlationId = "steer", but the current workflow already passes the dynamicctx.input.correlationId(vibersyn-process.tsx#L167-L174). The spec describes a bug that no longer exists β worth updating before someone "fixes" it back.routeRequest/ internal connection shapes. They work today, but can shift on any minor β Β§3 has the sanctioned replacement in flight.3 Β· Upstream gaps β fixes in flight on smithersai/smithers
Your embedding surfaced these. We're filing them from this review and will cross-link the numbers here as they land on the tracker.
WaitForEventtimeout deadlines don't wake a parked run (gateway.js@v0.31.0#L5852, deferred-state-bridge.js#L1135)Gateway.registertyping is non-generic (index.d.ts@v0.31.0#L1515)4 Β· Proposed sub-issues
claudeCLI lifecycle in the build backend β smithers.ts#L1-L14, #L70-L155 spawn the CLI directly and hand-roll timeout/salvage; running it as a smithers agent (timeoutMs/heartbeatTimeoutMs, durable retries) deletes that blockcorrelationIdcode (Β§2)Compiled against vibecode-room@355e89d with every claim checked against smithers v0.31.0 source. We're happy to send PRs β the version bump is the natural first one.