0.12.11

A message keeps its turn's thread; Google ADK streamed-text, Interactions and credential fixes

A patch release. All six packages move to 0.12.11 together. @silverprotocol/core and @silverprotocol/google-adk change. The spec revision stays 1.0.0-draft.8.

Who should upgrade

  • You run Google ADK 2.4.0 in StreamingMode.SSE with a planner that writes state. The package’s peer range (>=1.0.0 <3) admits 2.4.0, though the 0.12 line is verified at 2.2.1. Streamed text rendered twice on 0.12.10; it now renders once.
  • You run Google ADK on the Interactions route (useInteractionsApi) in StreamingMode.SSE. A response now renders once, where 0.12.10 rendered it again after turn.done.
  • Anyone else gets the same output as from 0.12.10: all 103 replayed runs are byte-identical (see Wire version). Read “Upgrading from 0.12.10” first: the message-thread change applies to every framework package’s output.

Upgrading from 0.12.10

  • A message lands on its turn’s thread. When a message.start names a threadId other than the one its turn already has (from the turn’s turn.start, from its parent’s thread for a nested turn, or from an earlier message.start), the reducer now creates the message on the turn’s thread. 0.12.10 kept the message.start’s own threadId. The turn record’s thread doesn’t move, and the fold doesn’t resync. A message.start for a turn that has no thread yet still gives the turn its thread. When a nested turn’s parent was never opened, the turn takes the thread of a turn the fold has already opened or, with none, the parent turn id its subagent.start names, as before; its messages now take the same thread.
  • Google ADK: an unmatched function response that holds credential material is no longer carried. A functionResponse that carries no id (or an empty one) and answers no pending call under its name rides ext.google.unparsed whole. When it holds an ADK credential object or a response to adk_request_credential, at any depth, the package now emits one non-terminal error event in its place, with code "adk_orphan_credential_response" and a fixed message, and nothing of the response rides (spec §8.0, graceful degradation). 0.12.10 carried such a response whole, credential members included. The error doesn’t close the turn. An unmatched response without credential material still rides whole, a reply named adk_request_credential keeps its allowlisted carry, and a response that answers a call still maps to tool.done. A credential an application keeps in a shape of its own isn’t recognised, and rides unchanged.

@silverprotocol/core

  • The message-thread change above. No other fold rule changes.

Google ADK

  • A planner’s state record no longer doubles streamed text. From @google/adk 2.4.0, with a planner that writes state, ADK yields an event that only records the state change inside a model call, between its streamed chunks and its repeat of them in StreamingMode.SSE. 0.12.10 read that event as the end of the call, so the repeat rendered the text a second time. On the 0.12 line the package matches streamed text per turn: the record still maps to its state.delta, and it holds the match open only when its own source streamed the match’s last chunk. While it holds, that source’s next chunk, partial: false repeat or another such record keeps the match, and an event from any other source, or one without a partial flag, ends it first, as before. A record of another source ends the match, as before. The repeat renders nothing, as without the record.
  • A repeated close of an Interactions response maps once. On the Interactions route (useInteractionsApi) in StreamingMode.SSE, any @google/adk from 2.2.1 through 2.4.0 closes a response at interaction.completed and again when the stream ends, and a completed status update can close it once more, each time with every part so far. 0.12.10 rendered the response again at the second close, after turn.done. A close that maps to exactly what the response’s previous close mapped to, its id and timestamp aside, with no other event of the response between them, now emits nothing. A close with any mapped difference maps as before, and a partial: false event ADK yields mid-response (a media part, a finished function call) is not a close and always maps.
  • The unmatched-response change under “Upgrading from 0.12.10”.

With @ai-sdk/openai 4.0.110 or later

  • If your host resolves @ai-sdk/openai 4.0.110 or later, an OpenAI Responses refusal reaches the stream. From 4.0.110, the provider streams a refusal as the assistant’s text and finishes the step with content-filter. This package maps that finish, as before, to turn.done with finishReason "safety_blocked", and sets no finishReasonRaw for it. With 4.0.109 or earlier, the refusal text never reached the stream: the host saw an empty assistant message, and the turn closed with finishReason "stop". The change is in the provider; this package’s code doesn’t change, and the 0.12 line’s verified versions don’t include 4.0.110.

The other four packages change only their version.

Verified versions

Unchanged from 0.12.10. Each framework package is verified end to end, tool calls included, against a live run at:

  • @anthropic-ai/claude-agent-sdk 0.3.292;
  • @openai/agents 0.21.0;
  • @google/adk 2.2.1;
  • ai 7.0.130.

Wire version

1.0.0-draft.8, unchanged. Replaying every recorded stream and the Google ADK pause fixtures in the release’s test corpus through the 0.12.10 and 0.12.11 packages, all 103 runs give byte-identical events. The replay compares the events each package emits, not a fold of them, and no recorded stream or fixture carries an unmatched credential response, a planner’s state record or a stream on the Interactions route; those changes rest on their own tests. A live capture on the Interactions route, recorded outside that corpus in StreamingMode.SSE at gemini-3.8-flash, shows the repeated-close fix: 0.12.10 renders the response twice, in 30 events, and 0.12.11 renders it once, in 19.

npm’s provenance for all six 0.12.11 packages attests typescript-sdk commit a0bf753e.