[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

@chat-adapter/shared@4.44.0

Minor Changes

• 17f6005: Add the @chat-adapter/shared/mentions subpath, which exports maskCodeSpans and replaceBareMentions without importing chat or any other runtime package. Runtime-free entry points, such as @chat-adapter/slack/webhook, can use the same code-span rules as the adapters.

Patch Changes

• Updated dependencies [b8b6f0f]
• Updated dependencies [49e6833]
  • chat@4.44.0

@chat-adapter/slack@4.44.0

Minor Changes

• 17f6005: Add isBotMention(event, botUserId), which reports whether a Slack message or app_mention event invokes the bot the same way the adapter sets message.isMention. Slack sends app_mention for the bot's token inside code, so the event type alone does not show that a user addressed the bot. Apps that route or acknowledge Slack events before a Chat SDK bot handles them can use it to make the same decision. It is exported from @chat-adapter/slack/webhook, which imports no runtime packages, and from @chat-adapter/slack, and follows the isBotMention contract documented for adapters. botUserId must be a Slack user ID; an empty or malformed value throws a TypeError.

• bb9c3dd: withBotToken accepts an optional botUserId, so a multi-workspace adapter can recognize the bot's own messages outside webhook handling. A multi-workspace adapter has no fixed bot user ID, and withBotToken previously stored only the token and installationId. Inside the callback, adapter.botUserId was undefined, and messages the bot had posted, such as those returned by fetchMessages, had author.isMe set to false. Pass the installation's botUserId from getInstallation:

  const install = await slackAdapter.getInstallation(teamId);
  if (!install) throw new Error("Workspace not installed");
  
  await slackAdapter.withBotToken(install.botToken, fn, {
    installationId: teamId,
    botUserId: install.botUserId,
  });

Patch Changes

• 1ba1cc4: Parse bulleted and numbered lists from Slack's message composer into mdast list nodes, including nested items. Slack flattens these lists in event.text to lines starting with • or ◦, so when a message's rich_text blocks contain a list, the adapter now builds the message body from the blocks. message.text for such a list reads one\ntwo instead of • one\n• two. Literal list text such as --- and [x] stays literal, and paragraphs in subsequent rich text blocks stay outside the list.

  Messages without a list, with blocks other than rich_text and tables, or with unsupported rich text elements still use event.text. This preserves references such as message_mention that the block renderer cannot reconstruct.

  Link labels that contain > in rich text blocks, including table cells, now stay inside the link instead of ending it early.

• Updated dependencies [b8b6f0f]

• Updated dependencies [49e6833]

• Updated dependencies [17f6005]

  • chat@4.44.0
  • @chat-adapter/shared@4.44.0

@chat-adapter/telegram@4.44.0

Minor Changes

• 8494511: Add a longMessages: "split" option that sends text posts longer than Telegram's message limit as consecutive messages instead of truncating them. Parts are paced to Telegram's per-chat limits, split tables repeat their header row, and a rate-limited part after the first is sent again after Telegram's retry_after period. Ordered lists in MarkdownV2 messages now keep their start number.

Patch Changes

• Updated dependencies [b8b6f0f]
• Updated dependencies [49e6833]
• Updated dependencies [17f6005]
  • chat@4.44.0
  • @chat-adapter/shared@4.44.0

chat@4.44.0

Minor Changes

• 49e6833: Add BaseFormatConverter.parseIncoming(), which returns a message's text and formatted AST from a single markdown parse. Very long text can be slow to parse, so text longer than maxIncomingParseLength (default DEFAULT_MAX_INCOMING_PARSE_LENGTH, 16,384 characters) skips parsing: text holds the raw input and formatted holds it as a single text node. Subclasses can override maxIncomingParseLength.

  The Discord, Google Chat, GitHub, GitLab, and Teams adapters now use parseIncoming() for incoming messages, so they parse each message once instead of twice. Messages over the limit keep their content but lose markdown structure in formatted. In Teams, text for such a message also keeps the raw HTML tags, and in GitLab a mention inside a code block or quote in such a note counts as a mention.

  Upstream releases fix slow parsing of long lists in mdast-util-from-markdown 2.1.0 and of large tables in micromark-extension-gfm-table 2.1.2. Both are within the ranges Chat SDK's dependencies already accept, so update them in your lockfile to pick up the fixes. Keep micromark at 4.0.2, since 4.0.3 slows down parsing of block quotes, nested lists, and setext headings.

Patch Changes

• b8b6f0f: requestApproval() from chat/workflow now verifies each click server-side instead of reading the decision and user from the webhook request. When a button with a callbackUrl is clicked, Chat SDK now stores a receipt in the state adapter with the verified action ID, user, thread, and message, and adds its identifier to the callback payload as receipt. requestApproval() takes the decision and user only from that receipt, accepts each receipt once, and ignores requests without a valid one. If the receipt can't be stored, Chat SDK skips the callback POST and restores the button's callback token so the user can click again. When someone outside approvers clicks, requestApproval() now also refreshes the card's buttons, because each button's callback works only once and that click previously left the button unusable for approvers.

  The Chat instance that handles the click and the one running the workflow must share a state adapter, and requestApproval() now throws when passed a Thread that wasn't created by Chat SDK. Errors logged when a callback URL POST fails now include only the URL's origin, under callbackOrigin instead of callbackUrl.

• @chat-adapter/catalog@4.44.0

@chat-adapter/discord@4.44.0

Patch Changes

• 49e6833: Add BaseFormatConverter.parseIncoming(), which returns a message's text and formatted AST from a single markdown parse. Very long text can be slow to parse, so text longer than maxIncomingParseLength (default DEFAULT_MAX_INCOMING_PARSE_LENGTH, 16,384 characters) skips parsing: text holds the raw input and formatted holds it as a single text node. Subclasses can override maxIncomingParseLength.

  The Discord, Google Chat, GitHub, GitLab, and Teams adapters now use parseIncoming() for incoming messages, so they parse each message once instead of twice. Messages over the limit keep their content but lose markdown structure in formatted. In Teams, text for such a message also keeps the raw HTML tags, and in GitLab a mention inside a code block or quote in such a note counts as a mention.

  Upstream releases fix slow parsing of long lists in mdast-util-from-markdown 2.1.0 and of large tables in micromark-extension-gfm-table 2.1.2. Both are within the ranges Chat SDK's dependencies already accept, so update them in your lockfile to pick up the fixes. Keep micromark at 4.0.2, since 4.0.3 slows down parsing of block quotes, nested lists, and setext headings.

• Updated dependencies [b8b6f0f]

• Updated dependencies [49e6833]

• Updated dependencies [17f6005]

  • chat@4.44.0
  • @chat-adapter/shared@4.44.0

@chat-adapter/gchat@4.44.0

Patch Changes

• 49e6833: Add BaseFormatConverter.parseIncoming(), which returns a message's text and formatted AST from a single markdown parse. Very long text can be slow to parse, so text longer than maxIncomingParseLength (default DEFAULT_MAX_INCOMING_PARSE_LENGTH, 16,384 characters) skips parsing: text holds the raw input and formatted holds it as a single text node. Subclasses can override maxIncomingParseLength.

  The Discord, Google Chat, GitHub, GitLab, and Teams adapters now use parseIncoming() for incoming messages, so they parse each message once instead of twice. Messages over the limit keep their content but lose markdown structure in formatted. In Teams, text for such a message also keeps the raw HTML tags, and in GitLab a mention inside a code block or quote in such a note counts as a mention.

  Upstream releases fix slow parsing of long lists in mdast-util-from-markdown 2.1.0 and of large tables in micromark-extension-gfm-table 2.1.2. Both are within the ranges Chat SDK's dependencies already accept, so update them in your lockfile to pick up the fixes. Keep micromark at 4.0.2, since 4.0.3 slows down parsing of block quotes, nested lists, and setext headings.

• Updated dependencies [b8b6f0f]

• Updated dependencies [49e6833]

• Updated dependencies [17f6005]

  • chat@4.44.0
  • @chat-adapter/shared@4.44.0

@chat-adapter/github@4.44.0

Patch Changes

• 49e6833: Add BaseFormatConverter.parseIncoming(), which returns a message's text and formatted AST from a single markdown parse. Very long text can be slow to parse, so text longer than maxIncomingParseLength (default DEFAULT_MAX_INCOMING_PARSE_LENGTH, 16,384 characters) skips parsing: text holds the raw input and formatted holds it as a single text node. Subclasses can override maxIncomingParseLength.

  The Discord, Google Chat, GitHub, GitLab, and Teams adapters now use parseIncoming() for incoming messages, so they parse each message once instead of twice. Messages over the limit keep their content but lose markdown structure in formatted. In Teams, text for such a message also keeps the raw HTML tags, and in GitLab a mention inside a code block or quote in such a note counts as a mention.

  Upstream releases fix slow parsing of long lists in mdast-util-from-markdown 2.1.0 and of large tables in micromark-extension-gfm-table 2.1.2. Both are within the ranges Chat SDK's dependencies already accept, so update them in your lockfile to pick up the fixes. Keep micromark at 4.0.2, since 4.0.3 slows down parsing of block quotes, nested lists, and setext headings.

• Updated dependencies [b8b6f0f]

• Updated dependencies [49e6833]

• Updated dependencies [17f6005]

  • chat@4.44.0
  • @chat-adapter/shared@4.44.0

@chat-adapter/gitlab@4.44.0

Patch Changes

• 49e6833: Add BaseFormatConverter.parseIncoming(), which returns a message's text and formatted AST from a single markdown parse. Very long text can be slow to parse, so text longer than maxIncomingParseLength (default DEFAULT_MAX_INCOMING_PARSE_LENGTH, 16,384 characters) skips parsing: text holds the raw input and formatted holds it as a single text node. Subclasses can override maxIncomingParseLength.

  The Discord, Google Chat, GitHub, GitLab, and Teams adapters now use parseIncoming() for incoming messages, so they parse each message once instead of twice. Messages over the limit keep their content but lose markdown structure in formatted. In Teams, text for such a message also keeps the raw HTML tags, and in GitLab a mention inside a code block or quote in such a note counts as a mention.

  Upstream releases fix slow parsing of long lists in mdast-util-from-markdown 2.1.0 and of large tables in micromark-extension-gfm-table 2.1.2. Both are within the ranges Chat SDK's dependencies already accept, so update them in your lockfile to pick up the fixes. Keep micromark at 4.0.2, since 4.0.3 slows down parsing of block quotes, nested lists, and setext headings.

• Updated dependencies [b8b6f0f]

• Updated dependencies [49e6833]

• Updated dependencies [17f6005]

  • chat@4.44.0
  • @chat-adapter/shared@4.44.0

@chat-adapter/gmail@4.44.0

Patch Changes

• Updated dependencies [b8b6f0f]
• Updated dependencies [49e6833]
• Updated dependencies [17f6005]
  • chat@4.44.0
  • @chat-adapter/shared@4.44.0

@chat-adapter/instagram@4.44.0

Patch Changes

• Updated dependencies [b8b6f0f]
• Updated dependencies [49e6833]
• Updated dependencies [17f6005]
  • chat@4.44.0
  • @chat-adapter/shared@4.44.0

@chat-adapter/linear@4.44.0

Patch Changes

• Updated dependencies [b8b6f0f]
• Updated dependencies [49e6833]
• Updated dependencies [17f6005]
  • chat@4.44.0
  • @chat-adapter/shared@4.44.0

@chat-adapter/messenger@4.44.0

Patch Changes

• Updated dependencies [b8b6f0f]
• Updated dependencies [49e6833]
• Updated dependencies [17f6005]
  • chat@4.44.0
  • @chat-adapter/shared@4.44.0

@chat-adapter/notion@4.44.0

Patch Changes

• 94a522c: Reject posts to a discussion or block that isn't on the page in the thread ID. Notion routes a reply by discussion_id and a new block discussion by parent.block_id without checking the page, so a thread ID such as notion:{pageA}:{discussionOnPageB} posted on page B while its channel was page A. postMessage now throws a ValidationError before uploading files or posting. Block targets are checked against the block's current page, and child page or child database blocks are rejected. Discussion targets must match a page the adapter recorded in the state adapter when a webhook arrived for the discussion, the bot started it, or it appeared in the page's comment list. That list only covers unresolved, page-level discussions, so an inline or block discussion with no record, such as a thread ID saved before this release, is rejected until a new comment in it arrives through the webhook.
• Updated dependencies [b8b6f0f]
• Updated dependencies [49e6833]
• Updated dependencies [17f6005]
  • chat@4.44.0
  • @chat-adapter/shared@4.44.0

@chat-adapter/teams@4.44.0

Patch Changes

• 49e6833: Add BaseFormatConverter.parseIncoming(), which returns a message's text and formatted AST from a single markdown parse. Very long text can be slow to parse, so text longer than maxIncomingParseLength (default DEFAULT_MAX_INCOMING_PARSE_LENGTH, 16,384 characters) skips parsing: text holds the raw input and formatted holds it as a single text node. Subclasses can override maxIncomingParseLength.

  The Discord, Google Chat, GitHub, GitLab, and Teams adapters now use parseIncoming() for incoming messages, so they parse each message once instead of twice. Messages over the limit keep their content but lose markdown structure in formatted. In Teams, text for such a message also keeps the raw HTML tags, and in GitLab a mention inside a code block or quote in such a note counts as a mention.

  Upstream releases fix slow parsing of long lists in mdast-util-from-markdown 2.1.0 and of large tables in micromark-extension-gfm-table 2.1.2. Both are within the ranges Chat SDK's dependencies already accept, so update them in your lockfile to pick up the fixes. Keep micromark at 4.0.2, since 4.0.3 slows down parsing of block quotes, nested lists, and setext headings.

• 5781d46: Set message.isMention to false when an incoming message's mention entities do not mention the bot. It was left unset, so Chat fell back to matching @userName in the text, and a bot named Bot also handled mentions of another bot named Bot Dev. The mention entity ID is now compared without case, and a plain-text @bot without a mention entity no longer counts as a mention in group chats or channels.

• Updated dependencies [b8b6f0f]

• Updated dependencies [49e6833]

• Updated dependencies [17f6005]

  • chat@4.44.0
  • @chat-adapter/shared@4.44.0

@chat-adapter/twilio@4.44.0

Patch Changes

• Updated dependencies [b8b6f0f]
• Updated dependencies [49e6833]
• Updated dependencies [17f6005]
  • chat@4.44.0
  • @chat-adapter/shared@4.44.0

@chat-adapter/twitch@4.44.0

Patch Changes

• Updated dependencies [b8b6f0f]
• Updated dependencies [49e6833]
• Updated dependencies [17f6005]
  • chat@4.44.0
  • @chat-adapter/shared@4.44.0

@chat-adapter/web@4.44.0

Patch Changes

• Updated dependencies [b8b6f0f]
• Updated dependencies [49e6833]
• Updated dependencies [17f6005]
  • chat@4.44.0
  • @chat-adapter/shared@4.44.0

@chat-adapter/whatsapp@4.44.0

Patch Changes

• Updated dependencies [b8b6f0f]
• Updated dependencies [49e6833]
• Updated dependencies [17f6005]
  • chat@4.44.0
  • @chat-adapter/shared@4.44.0

@chat-adapter/x@4.44.0

Patch Changes

• Updated dependencies [b8b6f0f]
• Updated dependencies [49e6833]
• Updated dependencies [17f6005]
  • chat@4.44.0
  • @chat-adapter/shared@4.44.0

@chat-adapter/state-ioredis@4.44.0

Patch Changes

• Updated dependencies [b8b6f0f]
• Updated dependencies [49e6833]
  • chat@4.44.0

@chat-adapter/state-memory@4.44.0

Patch Changes

• Updated dependencies [b8b6f0f]
• Updated dependencies [49e6833]
  • chat@4.44.0

@chat-adapter/state-pg@4.44.0

Patch Changes

• Updated dependencies [b8b6f0f]
• Updated dependencies [49e6833]
  • chat@4.44.0

@chat-adapter/state-redis@4.44.0

Patch Changes

• Updated dependencies [b8b6f0f]
• Updated dependencies [49e6833]
  • chat@4.44.0

@chat-adapter/catalog@4.44.0

@chat-adapter/tests@4.44.0

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated
chat	Ready	Preview, v0	Oct 27, 2026 2:50pm UTC