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

agents@0.18.0

Minor Changes

• #1860 f5b1dd8 Thanks @mattzcarey! - Group SDK-managed initialization, startup, chat interactions, turns, and durable submissions into semantic phases. Storage-heavy setup, hydration, recovery, request persistence, and response persistence each receive a named bucket, keeping inference and tool spans visible without discarding lower-level Durable Object SQLite spans. Each span records agent identity, storage phase, a stable marker for UI grouping, and operation-specific metadata. No-op on runtimes without the tracing API.

• #1860 f5b1dd8 Thanks @mattzcarey! - Add agents/observability/ai with wrapAISDK for AI SDK v6 and createAISDKTelemetry for AI SDK v7. Both project Cloudflare-native invoke_agent {agent}, chat {model}, and execute_tool {tool} spans using scalar OpenTelemetry GenAI attributes, including request settings, token usage, finish reasons, tool-call IDs, model-call time to first chunk, bounded AI SDK v6 approval lifecycle spans, and conditional AI Gateway log references. Payload storage is opt-in: storeMessages writes OTel-schema input/output message arrays to chat ({ role, parts }, canonical text/reasoning/tool parts, and output finish_reason; oldest messages are dropped past the budget while protecting the first two), and storeTools writes arguments/results to execute_tool. The flags themselves are never emitted as span attributes. Schemas, request headers, provider options, and raw errors remain excluded.

Patch Changes

• #1959 a3cbed1 Thanks @mattzcarey! - Cache MCP JSON Schema conversion for the current catalog on each live connection, and let Think agents skip direct MCP AI-tool exposure when those tools are exposed through Code Mode or another mechanism outside Think's automatic tool set.

• #1963 3ce98ff Thanks @mattzcarey! - Reconcile stale useAgentChat server-streaming state after an errored client reconnects.

  Reconnect probes now include correlation IDs, and STREAM_RESUME_NONE distinguishes globally idle agents from active continuations owned by another connection. The hook clears fallback streaming state only for a correlated idle response. Reconnect opens are retained while a prior resume or status transition settles, in-flight handshakes are retransmitted on replacement sockets, and all AI SDK resume entry points share one serialization gate.

• #1944 fff4131 Thanks @cjol! - Make the agents/vite turndown stub fail with a diagnostic error when app code calls it directly, and document the stubTurndown: false opt-out for applications that use turndown themselves.

• #1926 6861933 Thanks @mattzcarey! - Verify MCP OAuth callback state before changing a connection. Forged, expired, and replayed callbacks no longer disrupt the genuine authorization flow, and addMcpServer() refreshes auth URLs whose embedded state has expired.

• #1924 c19d58a Thanks @mattzcarey! - Harden MCP connection recovery: honor retry budgets for resolved connection failures, finish in-flight restore work before stable-id migration, and close connections replaced by the legacy connect() path.

• #1923 33e59c4 Thanks @mattzcarey! - Recover streamable HTTP connections when a server rejects a persisted session with HTTP 404. The client clears the stale session from memory and storage, initializes a new session, and rediscovers capabilities once.

@cloudflare/think@0.14.0

Minor Changes

• #1860 f5b1dd8 Thanks @mattzcarey! - Think agents emit Cloudflare-native traces out of the box with no tracing setup beyond enabling Worker traces. Named lifecycle and storage-phase spans group Durable Object internals around startup, request persistence, turn preparation, inference, result persistence, recovery, alarms, and durable submissions. Every inference call produces an invoke_agent {agent class} span with chat {model} and execute_tool {tool} children. Think supplies durable identity and turn metadata. Payload storage remains off by default; agents can set storeMessages for OTel-schema model messages ({ role, parts }, canonical tool parts, and output finish_reason) on chat and storeTools for tool arguments/results on execute_tool. These flags configure the wrapper and are never emitted as metadata or attributes. Model streams are finalized on early exit, and durable submissions run from their awaited alarm invocation so spans do not outlive invocation boundaries.

Patch Changes

• #1860 f5b1dd8 Thanks @mattzcarey! - Group SDK-managed initialization, startup, chat interactions, turns, and durable submissions into semantic phases. Storage-heavy setup, hydration, recovery, request persistence, and response persistence each receive a named bucket, keeping inference and tool spans visible without discarding lower-level Durable Object SQLite spans. Each span records agent identity, storage phase, a stable marker for UI grouping, and operation-specific metadata. No-op on runtimes without the tracing API.

• #1959 a3cbed1 Thanks @mattzcarey! - Cache MCP JSON Schema conversion for the current catalog on each live connection, and let Think agents skip direct MCP AI-tool exposure when those tools are exposed through Code Mode or another mechanism outside Think's automatic tool set.

• #1963 3ce98ff Thanks @mattzcarey! - Reconcile stale useAgentChat server-streaming state after an errored client reconnects.

  Reconnect probes now include correlation IDs, and STREAM_RESUME_NONE distinguishes globally idle agents from active continuations owned by another connection. The hook clears fallback streaming state only for a correlated idle response. Reconnect opens are retained while a prior resume or status transition settles, in-flight handshakes are retransmitted on replacement sockets, and all AI SDK resume entry points share one serialization gate.

• #1937 ed9f42b Thanks @cjol! - Forward routing props through the Think onStart lifecycle wrapper.

• #1860 f5b1dd8 Thanks @mattzcarey! - Run durable submissions only from their awaited scheduled drain, preventing model work and tracing spans from outliving the short submission-acceptance invocation.

• #1949 32e7390 Thanks @cjol! - Warn when getSkills() makes an overridden getSystemPrompt() fallback-only and document using Session context blocks for always-on instructions.

@cloudflare/ai-chat@0.9.4

Patch Changes

• #1860 f5b1dd8 Thanks @mattzcarey! - Group SDK-managed initialization, startup, chat interactions, turns, and durable submissions into semantic phases. Storage-heavy setup, hydration, recovery, request persistence, and response persistence each receive a named bucket, keeping inference and tool spans visible without discarding lower-level Durable Object SQLite spans. Each span records agent identity, storage phase, a stable marker for UI grouping, and operation-specific metadata. No-op on runtimes without the tracing API.

• #1963 3ce98ff Thanks @mattzcarey! - Reconcile stale useAgentChat server-streaming state after an errored client reconnects.

  Reconnect probes now include correlation IDs, and STREAM_RESUME_NONE distinguishes globally idle agents from active continuations owned by another connection. The hook clears fallback streaming state only for a correlated idle response. Reconnect opens are retained while a prior resume or status transition settles, in-flight handshakes are retransmitted on replacement sockets, and all AI SDK resume entry points share one serialization gate.

@cloudflare/codemode@0.4.4

Patch Changes

• #1969 80ad8de Thanks @mattzcarey! - Add framework-neutral execute, search, and describe methods to the durable Code Mode runtime handle so MCP servers and other non-AI-SDK hosts can invoke execution and discovery directly. Search and describe results now identify methods that require approval.

@cloudflare/shell@0.4.2

Patch Changes

• #1716 8ecdb61 Thanks @mattzcarey! - fix(shell): replace LIKE pattern matching with primary-key range scans in Workspace.rm({ recursive: true }) and the glob prefilter. D1 can reject the previous LIKE ? ESCAPE ? queries with D1_ERROR: LIKE or GLOB pattern too complex: SQLITE_ERROR; the range predicate (path >= '{dir}/' AND path < '{dir}0') avoids that limit, scans the path index directly, and needs no escaping of %/_ in path names.

@cloudflare/voice@0.3.5

Patch Changes

• #1912 219d59b Thanks @cjol! - Add AssemblyAI and ElevenLabs streaming STT providers for the voice pipeline.

@cloudflare/worker-bundler@0.2.2

Patch Changes

• #1961 365ced3 Thanks @agent-think! - fix: eliminate polynomial-time ReDoS in import/export matching. The importExportRegex in rewriteImports and the regex fallback in parseImports matched the import clause with [\w*{}\s,]+ followed by \s+, letting both quantifiers consume the same whitespace and backtrack catastrophically on near-match inputs (import + 10k spaces took ~175s). Clauses are now matched as non-whitespace tokens separated by whitespace and bounded at the next import/export keyword, eliminating both overlapping backtracking and repeated suffix scans while preserving valid imports.

@cloudflare/agent-think@0.0.2

Patch Changes

• Updated dependencies [f5b1dd8, f5b1dd8, a3cbed1, 3ce98ff, ed9f42b, fff4131, 6861933, c19d58a, 33e59c4, f5b1dd8, f5b1dd8, 32e7390]:
  • agents@0.18.0
  • @cloudflare/ai-chat@0.9.4
  • @cloudflare/think@0.14.0

@cloudflare/voice-assemblyai@0.0.2

Patch Changes

• #1912 219d59b Thanks @cjol! - Add AssemblyAI and ElevenLabs streaming STT providers for the voice pipeline.

@cloudflare/voice-elevenlabs@0.0.2

Patch Changes

• #1912 219d59b Thanks @cjol! - Add AssemblyAI and ElevenLabs streaming STT providers for the voice pipeline.

[devin-ai-integration]

✅ Devin Review: No Issues Found

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