cloudflare/agents @cloudflare/think@0.20.0

on GitHub

Minor Changes

• #2388 1031c0f Thanks @threepointone! - Require ai@^7 and @ai-sdk/react@^4, because Think bundles providers that return AI SDK v7 models. See Think dependencies.

• #2056 20142e5 Thanks @AntoniTok! - Breaking: declared scheduled tasks now run on the root agent only; override getScheduledTasksScope() to return "all" for per-facet scheduling. See Think scheduled tasks.

• #2330 e80278e Thanks @threepointone! - Browser chat turns now use the web channel, so its configured tools and turn limit apply. See Think channels.

Patch Changes

• #2390 c55ec80 Thanks @threepointone! - Report a failed agent-tool child as failed even when it was evicted before recording the failure. See Agent tools.

• #2384 f904999 Thanks @threepointone! - Fix agent-tool chunks being duplicated or dropped on reconnect, child re-attach, and fiber recovery. See Agent tools.

• #2364 5e0507e Thanks @threepointone! - Add eventDelivery: "terminal" to runAgentTool to forward only lifecycle, progress, and milestone events for a run. See Agent tools.

• #2334 7f564e7 Thanks @threepointone! - AI Chat and Think send the terminal done frame after persisting and broadcasting the assistant reply, so later sends are not overwritten. See Chat agents.

• #2391 d3fe93c Thanks @threepointone! - Keep a tool call's input when its approval request arrives before the input finishes streaming (#1872). See Human in the loop.

• #2326 a3e4eef Thanks @threepointone! - Fix two Think warnings that fired, or failed to fire, because of a false method-override check. See Think lifecycle hooks.

• #2408 6c8f85a Thanks @threepointone! - Require agents@>=0.25.0 so Think and AI Chat can import the chat helpers provided by that release.

• #2378 dbf170c Thanks @threepointone! - useAgentChat gains onTurnEnd, and terminal chat frames carry outcomes and user message IDs so clients can settle sends correctly across recovery and reconnects. Replay, tool callbacks, and turn state now remain consistent through terminal delivery; see Chat agents.

• #2005 c2f7672 Thanks @cjol! - Native RPC calls to async Agent and Think methods now start lifecycle initialization first; address Agents by name because raw IDs from newUniqueId() and idFromString() now fail their first async RPC. See Lifecycle.

• #2036 040db5c Thanks @justanotherbyte! - Stop storing compaction summaries that clients echo back, and hide duplicates already stored (#1984). See Sessions.

• #2405 11f87b5 Thanks @threepointone! - Context blocks can set whenChanged: "remind" to send changes as a reminder after the cached prompt instead of rewriting it. See Context.

• #2393 a7d0e02 Thanks @threepointone! - Continuation turns that fail before streaming now report an error to clients and onChatResponse instead of appearing to hang (#2381). See Client tools.

• #2323 57e2e46 Thanks @threepointone! - Messenger attachments load when fetchData returns a plain ArrayBuffer. See Messengers.

• #2325 6289a15 Thanks @threepointone! - Messenger replies now see every message in a quick burst, not just the last one. See Messengers.

• #2324 34df44b Thanks @threepointone! - Messenger replies no longer post a ... placeholder on adapters without native streaming. See Messengers.

• #2347 00a07bd Thanks @threepointone! - A recovered messenger turn delivers its answer to the thread instead of an "interrupted" apology. See Messengers.

• #2353 06d50eb Thanks @threepointone! - rejectExecution() accepts { autoContinue: false } to record a rejection without starting another turn. See Actions.

• #2238 c3a4010 Thanks @mattzcarey! - Interfaces can now be used as Props on Agent, AIChatAgent, Think, and routing helpers. See Routing.

• #2040 2afe0e0 Thanks @AntoniTok! - Keep assistant messages and tool outputs attached to the correct turn when a provider reuses a toolCallId; resolveToolMergeId is deprecated in favour of reconcileMessages. See Chat agents.

• #2039 c658b83 Thanks @justanotherbyte! - useAgentChat keeps messages sent while disconnected when reconnecting, including Think transcripts, instead of erasing them (#1983). See Chat agents.

• #2348 39361fa Thanks @threepointone! - Fix text appearing twice when reconnecting during a tool continuation. See Resumable streaming.

• #2340 3b278b2 Thanks @threepointone! - onChatResponse still fires, with recovered: true, for a turn persisted just before a Durable Object reset (#2266, #1842). See Think lifecycle hooks.

• #2404 9d125c8 Thanks @threepointone! - Add session.mirror() to keep an in-memory transcript in sync with a session's change feed. See Sessions.

• #2392 4f26402 Thanks @threepointone! - Settle approved tool calls that never ran once the conversation moves past them, so later turns don't send unresolved tool calls (#2382). See Human in the loop.

• #2321 7d58ade Thanks @threepointone! - Think can write server-authored messageMetadata onto assistant messages, per agent or per turn, and recovery preserves metadata even when keys match built-in object properties. See Think lifecycle hooks.

• #2329 558424c Thanks @threepointone! - Recovery continuations extend the interrupted assistant message instead of adding a second one. See Think.

• #2331 0c5fd95 Thanks @threepointone! - Resolving a paused execution removes the model's now-stale "waiting for approval" text from the transcript (#2054). See Actions.

• #2355 ccca389 Thanks @threepointone! - Add getGateway(model) to set AI Gateway options such as metadata and caching for string models (#2262). See Think.

• #2388 1031c0f Thanks @threepointone! - resolveModel() rejects malformed model ids up front, and @hf/... ids get the same Workers AI settings as @cf/.... See Think.

• #2354 d3cb4e1 Thanks @threepointone! - Add messengerConcurrency to choose the Chat SDK concurrency strategy for messenger replies (#2313). See Messengers.

• #2387 9f70bc8 Thanks @threepointone! - Messenger replies hold the thread lock and retain queued follow-ups during slow delivery; typing updates are serialized and settle before reply text arrives. See Think messengers.

• #2413 bcb1f12 Thanks @threepointone! - Fix regenerate continuing the old response instead of replacing it (#2028). See Multi-chat.

• #2417 9906881 Thanks @threepointone! - A regeneration interrupted by a restart now recovers on its own branch. See Multi-chat.

• #2356 f92420f Thanks @threepointone! - Older-message truncation moves in steps (truncationStep), so the prompt-cache prefix stays stable across turns (#2200). See Think.

• #2328 84e6e5c Thanks @threepointone! - Stream-stall recovery retries stalls before the first chunk and now calls onChatRecovery. See Think lifecycle hooks.

• #2362 7a30c1f Thanks @threepointone! - step.prompt() turns interrupted mid-stream now recover with their structured output instead of failing (#1727). See Workflows.

• #2351 0b13fe0 Thanks @threepointone! - Add waitForSubmission() for durable submissions and harden cancellation, status-hook, reset-turn, and paused-tool lifecycle handling so waits settle and terminal status is reported once. See Programmatic submissions.

• #2414 484f7ab Thanks @threepointone! - telegramMessenger adds a nativeStreaming option and works with verifyWebhook on @chat-adapter/telegram 4.38 (#2397). See Messengers.

• #2350 52e624e Thanks @threepointone! - Think hooks and tools can read the running turn identity, and activeChannel stays scoped to its owning turn. See Lifecycle hooks.

• #2343 a2f6f94 Thanks @threepointone! - Harden transient chat recovery across Think and AI Chat, including retry budgeting, cancellation, terminal error delivery, and recovery after restarts. See the agents/chat recovery helpers.

• #2339 1edc989 Thanks @threepointone! - Truncate older tool results after model conversion so toModelOutput and provider-executed tools keep valid outputs (#2014). See Sessions.

• #2290 5848d71 Thanks @ben-reitz! - runTurn({ mode: "wait" }) returns the assistant message from its own turn, even if other turns persist in the meantime. See Programmatic submissions.

• #2341 caef62a Thanks @threepointone! - Wait-mode TurnResult.output returns the parsed structured output, and submission inspections include the persisted messageId (#2263, #2264). See Programmatic submissions.