Version Packages

[github-actions]

This PR was opened by the Changesets release GitHub action. When you're ready to do a release, you can merge this and the packages will be published to npm automatically. If you're not ready to do a release yet, that's fine, whenever you add more changesets to main, this PR will be updated.

Releases

@cloudflare/think@0.6.0

Minor Changes

• #1456 787e73d Thanks @mattzcarey! - Stop applying pruneMessages({ toolCalls: "before-last-2-messages" }) to the model context by default. The previous default silently stripped client-side tool results (no execute, output supplied via addToolOutput) from any turn beyond the second, breaking multi-turn flows where the user's choices live in those tool results (see #1455). truncateOlderMessages still runs as before, so context cost stays bounded.

  This is a behavior change. Subclasses that relied on the old aggressive pruning can opt back in from beforeTurn:

  import { pruneMessages } from "ai";
  
  beforeTurn(ctx) {
    return {
      messages: pruneMessages({
        messages: ctx.messages,
        toolCalls: "before-last-2-messages"
      })
    };
  }

Patch Changes

• #1463 ab2b1db Thanks @whoiskatrin! - Avoid throwing when chat stream resume negotiation/replay races with a closed WebSocket connection. Resume protocol sends and the _handleStreamResumeAck fallback now go through sendIfOpen helpers that swallow the TypeError: WebSocket send() after close race instead of letting it propagate up through onMessage.

@cloudflare/voice@0.2.0

Minor Changes

• #1478 2c7d91b Thanks @whoiskatrin! - Add an enabled option to useVoiceAgent so React apps can delay creating and connecting a VoiceClient until async prerequisites such as capability tokens are ready.

Patch Changes

• #1458 84cb429 Thanks @whoiskatrin! - Fix Workers AI STT session edge cases for Flux and Nova 3.

  Flux now preserves the latest non-empty turn transcript from turn lifecycle events so an EndOfTurn event with an empty transcript can still emit the completed utterance. Flux StartOfTurn also drives server-side barge-in so model-detected user speech aborts active LLM/TTS playback promptly. Nova 3 now defensively normalizes finalized segment state before reading it to avoid stale teardown messages throwing during abnormal close paths.

• #1462 5f6214d Thanks @whoiskatrin! - Fix withVoice text streaming for AI SDK textStream responses so TTS audio is produced when onTurn() returns streamText(...).textStream directly.

agents@0.12.4

Patch Changes

• #1376 6561a3f Thanks @hrushikeshdeshpande! - Avoid throwing when chat stream resume negotiation/replay races with a closed WebSocket connection.

• #1476 3c48858 Thanks @whoiskatrin! - Fixed a bug that could cause client state to drift from internal Durable Object state when agent tool calls spanned a Durable Object restart. Recovery now defers user finish hooks until after agent startup and isolates hook failures so one failed mirror write does not block other recovered runs from finalizing.

@cloudflare/ai-chat@0.6.3

Patch Changes

• #1376 6561a3f Thanks @hrushikeshdeshpande! - Avoid throwing when chat stream resume negotiation/replay races with a closed WebSocket connection.

• #1487 752e25a Thanks @whoiskatrin! - Close the original WebSocket chat transport stream when the socket disconnects before a terminal response, preventing recovered chat continuations from leaving useAgentChat stuck in streaming state.

@cloudflare/codemode@0.3.5

Patch Changes

• #1468 186a2a4 Thanks @mattzcarey! - Add a browser-safe codemode export with an iframe sandbox executor and browser tool helper.

• #1470 1033fa2 Thanks @mattzcarey! - Resolve OpenAPI specs inside the codemode sandbox to avoid Worker Loader RPC size limits for heavily-referenced specs.

[devin-ai-integration]

✅ Devin Review: No Issues Found

Devin Review analyzed this PR and found no bugs or issues to report.