mirror of
https://github.com/vercel/chat.git
synced 2026-09-14 18:32:29 +08:00
@chat-adapter/x@4.34.0
487 Commits
| Author | SHA1 | Message | Date | |
|---|---|---|---|---|
 github-actions[bot]
|
f84b5911e7 |
chore(release): version packages (#695)
This PR was opened by the [Changesets release](https://github.com/changesets/action) 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/discord@4.34.0 ### Minor Changes - |
||
|
Remon Oldenbeuving
|
5c926f1987 |
fix(chat): preserve markdown whitespace in plain text (#604)
## Problem
When I mention a Chat SDK-powered bot on GitHub like this:
```
@bot
hi there!
```
The mention doesn't trigger any handling, because the `mdastToString`
helper replaces all newlines with an empty string, so the logs look like
this:
```
[chat-sdk] Checking message patterns { patternCount: 0, patterns: [], messageText: '@bothi there' }
[chat-sdk] No handlers matched message {
threadId: 'github:RSO/chat:issue:29',
text: '@bothi there'
}
```
## Summary
- Preserve structural markdown whitespace when extracting normalized
plain text from mdast.
- Add regression coverage for newline-separated bot mentions in core
markdown extraction, GitHub issue/review comments, and Chat mention
routing.
- Add a patch changeset for the behavior fix.
## Testing
- `pnpm check`
- `pnpm knip`
- `pnpm test:workspace`
- `pnpm --filter chat test`
- `pnpm --filter @chat-adapter/github test`
- `pnpm --filter chat typecheck`
- `pnpm --filter @chat-adapter/github typecheck`
## Notes
- `pnpm validate` was attempted and reached the full Turbo test graph,
but failed on `packages/integration-tests/src/replay-discord.test.ts`
(`should skip bot's own messages in subscribed threads`). Rerunning
`pnpm --filter @chat-adapter/integration-tests test --
src/replay-discord.test.ts` passed, so this appears unrelated to the
markdown whitespace change.
- `.github/CONTRIBUTING.md` requires signed commits. This environment
has no `gpg` binary and no SSH signing identities loaded, so the commit
in this PR is currently unsigned and may need to be re-signed before
merge.
---------
Signed-off-by: dancer <josh@afterima.ge>
Co-authored-by: dancer <josh@afterima.ge>
|
||
|
C. T. Lin
|
6714efc3a1 |
feat: support AI SDK v7 (ai@7) as a peer dependency (#691)
Closes #690 ## What Widens the AI SDK peer dependency ranges so the Chat SDK installs cleanly next to `ai@7`: - `chat`: `ai@^6.0.182 || ^7.0.0` - `@chat-adapter/web`: `ai@^6 || ^7`, `@ai-sdk/react@^3 || ^4`, `@ai-sdk/svelte@^4 || ^5`, `@ai-sdk/vue@^3 || ^4` This also unbreaks `create-chat-sdk` scaffolds, which install `ai@latest` (now v7) next to `chat` and currently hit a peer conflict out of the box. ## The one real v6 → v7 break In v7, `tool()` with an `execute` function returns `ExecutableTool<Tool<...>>` — an internal type from `@ai-sdk/provider-utils` that `ai` does not re-export. The `chat/ai` tool factories relied on inference, so declaration emit failed with TS2742 (17 errors). The factories now declare explicit `Tool<Input, Output>` return types, which is exactly the shape the previously published `.d.ts` already had — the public type surface is unchanged, and the emitted declarations only reference types from `ai` (portable for consumers on either major). Everything else checked out compatible: - v7 stream parts keep `text-delta` / `finish-step` shapes, so `fromFullStream` duck-typing works unchanged; `fullStream` remains as a deprecated alias - tool-level `needsApproval` is deprecated in v7 but still typed and honored - `createUIMessageStream`, `createUIMessageStreamResponse`, `isTextUIPart`, `UIMessage`, `UIMessageStreamWriter`, `ChatInit`, `DefaultChatTransport` all still exported — `@chat-adapter/web` needed zero source changes ## Other changes - devDependencies move to v7 so the workspace develops/tests against the latest major - `examples/nextjs-chat` and `examples/nuxt-chat` move to `ai@^7` (required — mixing majors across the workspace fails typecheck, since `chat`'s d.ts resolves `ai` types from its own devDependency) - Test-only: the `ToolExecutionOptions` stub type is now derived from `Tool["execute"]` because v7 made the generic parameter required - Changeset included (minor for `chat` and `@chat-adapter/web`) ## Verification The same source was verified against **both majors** (`ai@6.0.182` and `ai@7.0.17`): `tsc --noEmit` and the full test suites (`chat`: 1035 tests, `@chat-adapter/web`: 21 tests) pass on each. `pnpm validate` (knip + check + typecheck + test + build, including both examples) is green on v7. Note for adopters: `ai@7` itself requires Node.js ≥ 22 and is ESM-only; `chat` keeps `engines.node >= 20` since `ai` is an optional peer. 🤖 Generated with [Claude Code](https://claude.com/claude-code) Signed-off-by: chentsulin <chentsulin@gmail.com> Co-authored-by: Claude Fable 5 <noreply@anthropic.com> Co-authored-by: dancer <josh@afterima.ge> |
||
|
Shkumbin Hasani
|
2338a66544 |
feat(whatsapp): add sendTemplate for pre-approved template messages (#588)
Closes #585 ## Summary Adds `sendTemplate()` to the WhatsApp adapter for sending pre-approved [Message Templates](https://developers.facebook.com/docs/whatsapp/cloud-api/guides/send-message-templates) — the only message type the Cloud API accepts outside the 24-hour customer service window, and therefore required for business-initiated conversations (notifications, reminders, re-engagement). The inbound half already existed (`handleButtonResponse` dispatches template quick-reply taps to `onAction` handlers); this completes the outbound side. Notably, the package's `AGENTS.md` already documented `sendTemplate` and `WhatsAppTemplateMessage` as part of the public surface — this PR implements exactly that documented API. ## Changes - **`sendTemplate(threadId, template)`** on `WhatsAppAdapter` — posts a `type: "template"` payload through the existing `graphApiRequest` path and returns a `RawMessage`, mirroring `sendInteractiveMessage`/`sendSingleTextMessage` - **Types**: `WhatsAppTemplateMessage`, `WhatsAppTemplateComponent`, `WhatsAppTemplateParameter`, `WhatsAppTemplateButtonParameter` in `types.ts`, modeled on the Cloud API template object (header/body/button components; text, currency, date_time, and media parameters), re-exported from the package entry point - Templates are kept out of the `PostableMessage`/mdast pipeline — they're sent by name + variable components, not free-form markdown, and the adapter intentionally does not auto-substitute templates for outbound text posts (per AGENTS.md) - `openDM()` JSDoc now links to `sendTemplate` for the business-initiated path ## Docs - New "Template messages" section in the adapter README and `apps/docs/content/adapters/official/whatsapp.mdx` with a usage example - Added "Template messages" row to the README feature table - Corrected the feature-matrix frontmatter labels: `cardFormat` "WhatsApp templates" → "Interactive messages" and `fields` "Template variables" → "Formatted text" — cards render as Cloud API interactive messages (as the page body already states), and the old labels would now wrongly imply cards go through the new template API ## Usage ```typescript const threadId = await adapter.openDM("15551234567"); await adapter.sendTemplate(threadId, { name: "appointment_reminder", language: "en", components: [ { type: "body", parameters: [{ type: "text", text: "Tomorrow at 2pm" }], }, ], }); ``` ## Testing - 5 new tests in `index.test.ts` following the existing `fetch`-spy pattern: payload shape (name/language/`to`), component pass-through (body + URL button), empty-components omission, missing-message-ID error, and invalid thread ID rejection - `pnpm validate` (knip, check, typecheck, test, build) passes — 117/117 adapter tests green Includes a `minor` changeset for `@chat-adapter/whatsapp`. --------- Signed-off-by: dancer <josh@afterima.ge> Co-authored-by: dancer <josh@afterima.ge> |
||
|
josh
|
57fa85ca54 |
test(telegram): raise rich-truncate test timeout to 15s (#703)
## summary the `rich markdown messages truncate at the rich message limit` test runs ~5.5s in CI (coverage + shared runner) and intermittently exceeds the 5s default, failing unrelated PRs. raise its timeout to 15s ## test plan - `pnpm --filter @chat-adapter/telegram test` Signed-off-by: dancer <josh@afterima.ge> |
||
|
Ben Sabic
|
0f743c9b33 |
feat(slack): support native Slack agents (#698)
Builds on the Agent messaging experience support from #684 with a declarative config layer for building Slack agents, plus hardening for native streaming. Everything is configured on `createSlackAdapter()` — no per-event wiring required. ## Slack adapter (`@chat-adapter/slack`) ### `suggestedPrompts` Static payload or per-thread resolver, applied automatically when an assistant/agent thread opens: - `assistant_thread_started` (legacy `assistant_view`), with the thread's `thread_ts` - Messages-tab `app_home_opened` (with `agentView` enabled), without `thread_ts` so prompts pin atop the agent conversation The resolver receives the thread context (`channelId`, `userId`, legacy `threadTs`/`teamId`/`enterpriseId`, and normalized active-view `entities` under `agentView`); returning `null`/`undefined` skips the thread. Prompts beyond Slack's 4-prompt limit are dropped with a warning. Resolver/API failures are logged, never a webhook 500. Applied via `waitUntil` inside the request scope so multi-workspace token context propagates. ### `loadingMessages` Default rotating status strings for the assistant thinking indicator, used by `startTyping` and `setAssistantStatus` when no explicit status/messages are passed. ### `nativeStreaming` + automatic post-and-edit fallback - New `nativeStreaming` config (default `true`). Set `false` on Slack flavours without the `chat.startStream` family (e.g., GovSlack) to always stream via post-and-edit. - If the workspace rejects the **first** native streaming call, `stream()` falls back to throttled post-and-edit mid-stream instead of failing the reply; already-consumed text is preserved (it lives in the renderer). Permanent platform errors (`unknown_method`, `method_deprecated`, `feature_not_enabled`) latch native streaming off for subsequent streams on the adapter instance; transient errors don't latch. - Structured chunks (`task_update`/`plan_update`) are skipped in fallback mode; failures after native content has rendered still propagate (mixing surfaces would duplicate output). - Also updates the stale streaming description in the package AGENTS.md (the adapter now streams via `chat.startStream`/`appendStream`/`stopStream`, not `chat.update`). ### `feedbackButtons` Appends Slack's native thumbs up/down (a `context_actions` block with a `feedback_buttons` element) to every streamed reply on `chat.stopStream`, after any `StreamingPlan` `endWith` blocks. Pass `true` for defaults or an options object (`actionId`, labels, values). Clicks dispatch through the regular `block_actions` flow to `bot.onAction` with a positive/negative value — no new plumbing. Exports `buildFeedbackButtonsBlock(options?)` for attaching the same block to non-streamed messages. New exported types: `SlackFeedbackButtonsOptions`, `SlackSuggestedPrompt`, `SlackSuggestedPrompts`, `SlackSuggestedPromptsContext`, `SlackSuggestedPromptsOptions`. ## Docs - Configuration table rows for `agentView`, `suggestedPrompts`, `loadingMessages`, `nativeStreaming`, `feedbackButtons`. - New "Native streaming" and "Feedback buttons" sections plus declarative suggested-prompts examples. - All agent content grouped under a new **Advanced → Agents** subsection (Agent messaging experience → Assistants API → Native streaming → Feedback buttons). Heading titles unchanged, so existing anchors keep resolving. - TypeTable descriptions rewritten as plain text (they don't render markdown). ## Example app (`examples/nextjs-chat`) - `SLACK_AGENT_OPTIONS` shared across both Slack adapter branches: active-view-aware `suggestedPrompts` resolver, `loadingMessages`, `feedbackButtons` with an `ai_feedback` acknowledgment handler. - Env toggles: `SLACK_AGENT_VIEW` (agent_view mode) and `SLACK_NATIVE_STREAMING` (compare native vs post-and-edit). - Commented `agent_view` blocks in `slack-manifest.yml` (feature block, `assistant:write` scope, agent events) with a note that the switch is irreversible. - AI flows call `startTyping()` without an explicit status so configured loading messages rotate. ## Test plan - `pnpm validate` and `pnpm konsistent` pass. - 26 new unit tests: suggested prompts (static/resolver/agent_view/truncation/error paths), loading message defaults, native streaming fallback (opt-out, mid-stream fallback, permanent-error latching, transient non-latching, propagation after native render, structured-chunk skipping), feedback buttons (block shape, custom options, ordering after `endWith`, webhook round-trip of a click). - Each adapter commit was built and verified independently (typecheck + full suite green at every step) for bisectability. - Verified manually against a live `agent_view` workspace: prompts pinned on thread open, loading messages rotating in the thinking indicator, native token-by-token streaming in DMs and channel threads, post-and-edit fallback via the opt-out flag, and feedback clicks dispatching to `onAction`. ## Notes - One changeset covers the three adapter features (`minor` for `@chat-adapter/slack`). - Known follow-up (not in this PR): under `agentView`, a subscribed conversation-scoped DM thread (the #684 openDM bridge) routes DM messages to a thread without `thread_ts`, which silently pins DMs to post-and-edit. Worth deciding whether the bridge should keep per-message threading for replies or log loudly when it redirects. ## Screenshots | Suggested Prompts | Feedback Buttons | | --- | --- | | <img width="647" height="347" alt="CleanShot 2026-07-13 at 13 26 27" src="https://github.com/user-attachments/assets/4c89932b-ed19-4b4a-83ae-d3d022d0c120" /> | <img width="825" height="276" alt="CleanShot 2026-07-13 at 13 28 23" src="https://github.com/user-attachments/assets/6bdf6979-5209-4bed-b0bc-dfbee7165455" /> | --------- Signed-off-by: Ben Sabic <bensabic@users.noreply.github.com> Co-authored-by: Ben Sabic <bensabic@users.noreply.github.com> |
||
|
Dima Grossman
|
8bd8a57518 |
feat(whatsapp): send outbound files and attachments via Cloud API (#537)
Implement media upload, MIME mapping, caption fallbacks, and card+file sequencing. ## Summary The WhatsApp adapter previously ignored `files` and `attachments` on outbound `post()` calls (only text and interactive cards were sent). This PR implements full outbound media support via the [WhatsApp Cloud API](https://developers.facebook.com/docs/whatsapp/cloud-api/reference/media): 1. **Binary upload** — `POST /{phoneNumberId}/media` → `media_id` → typed media message 2. **Link passthrough** — HTTPS `Attachment.url` sent directly (no upload) 3. **Multi-file** — one WhatsApp message per file/attachment, sent sequentially 4. **Captions** — markdown or card fallback text on the first media message when supported 5. **Card + files** — media first, then interactive buttons (when applicable) **Packages:** `@chat-adapter/whatsapp` (minor) --- ## Supported inputs | Input | Description | |-------|-------------| | `files: FileUpload[]` | Binary buffers/blobs with `filename` and optional `mimeType`. Always uploaded via `/media`. | | `attachments: Attachment[]` | Typed media (`image` \| `file` \| `video` \| `audio`). Binary via `data` / `fetchData`, or HTTPS URL-only via `url`. | Both can be combined on `{ markdown }`, `{ raw }`, `{ ast }`, or `{ card }` postables. `files` are processed first, then `attachments`. ### Examples ```typescript // PDF with caption await thread.post({ markdown: "Here's the report", files: [{ data: pdfBuffer, filename: "report.pdf", mimeType: "application/pdf" }], }); // Multiple files (N sequential messages) await thread.post({ markdown: "Two files attached", files: [ { data: buf1, filename: "a.pdf", mimeType: "application/pdf" }, { data: buf2, filename: "b.png", mimeType: "image/png" }, ], }); // Card with buttons + image file await thread.post({ card: approvalCard, files: [{ data: proofBuffer, filename: "proof.png", mimeType: "image/png" }], }); // Files only (no text) await thread.post({ markdown: "", files: [{ data: buffer, filename: "data.xlsx" }], }); ``` --- ## Behavior reference ### Message flow (with media) ``` postMessage() ├─ files or attachments present? │ YES → postMessageWithMedia() │ ├─ Resolve text (card fallback OR markdown/raw/ast) │ ├─ Caption strategy (see below) │ ├─ For each file/attachment: upload (if binary) → sendMediaMessage() │ └─ Card present? │ ├─ interactive buttons → sendInteractiveMessage() │ └─ text-only card fallback → sendTextMessage() (if caption didn't already send text) │ └─ NO → existing text / card-only path (unchanged) ``` ### Multi-file WhatsApp allows **one media object per API message**. Multiple `files` or `attachments` in a single `post()` produce **N sequential messages**. The returned `RawMessage` is the **last** one sent (same convention as long-text chunking). | File index | Caption | |------------|---------| | First | Markdown / card fallback text (when caption rules allow) | | 2…N | No caption | ### Caption placement | Condition | Behavior | |-----------|----------| | Text ≤ 1024 chars, first media is not `audio`, media supports captions | Text sent as **caption** on first media message | | Text > 1024 chars | **Separate text message first**, then media with no captions | | First media is `audio` | **Separate text message first** (audio does not support captions), then audio | | No text (`markdown: ""`, files only) | Media only, no caption | ### MIME type → WhatsApp message type | MIME | WhatsApp `type` | |------|-----------------| | `image/jpeg`, `image/png` | `image` | | Other `image/*` (e.g. GIF, WebP, SVG) | `document` | | `video/mp4`, `video/3gpp` | `video` | | `audio/*` | `audio` | | Everything else (PDF, XLSX, etc.) | `document` | For `Attachment` without `mimeType`, the adapter uses `attachment.type` (`image` → image, `file` → document, etc.), then applies MIME rules when `mimeType` is set. ### Size limits (pre-flight) Throws `ValidationError` when binary size is known (before upload): | Type | Limit | |------|-------| | `image` | 5 MB | | `audio` | 16 MB | | `video` | 16 MB | | `document` | 100 MB | URL-only attachments skip size validation unless `attachment.size` is provided. ### Card + files When both a **card** and **files/attachments** are present: 1. **Media message(s)** first — caption uses `cardToFallbackText(card)` on the first media item 2. **Card message** second: - Valid reply buttons (1–3) → interactive button message - Otherwise → text fallback message (skipped if text was already sent as a leading message) ### Card image vs `files` (important) | How image is provided | Result | |-----------------------|--------| | `<Image>` child or `card.imageUrl` only (no `files`) | **No real image media.** Card becomes interactive text or text fallback; image URL may appear as plain text in fallback. | | `files` / `attachments` + card with buttons | **Real image message** + separate interactive button message | To send a photo with buttons, pass the image via `files` or `attachments`, not only as a card image child. ### Link passthrough - `Attachment` with **only** `url` (no `data` / `fetchData`) → `{ link: url }` in the media payload - URL **must** be `https://` - No `/media` upload call ### Binary resolution | Source | Path | |--------|------| | `FileUpload.data` | `toBuffer()` → `uploadMedia()` → `{ id }` | | `Attachment.data` / `fetchData` | Same | | `Attachment.url` only | `{ link }` passthrough | --- ## Out of scope (follow-ups) - Stickers (WebP encoding requirements) - Voice notes (`voice` vs `audio` distinction) - Media ID caching across posts (30-day expiry) - Interactive message **image headers** (card-embedded images without `files`) - Replay integration test mock extensions for `/media` - Edit/replace flows that include files --------- Co-authored-by: Ben Sabic <bensabic@users.noreply.github.com> |
||
|
Takuma Shibuya
|
2531a4227e |
fix username regexp (#621)
## Summary <!-- What does this PR do? --> Fix `detectMention` falsely matching `@bot` when `@bot-dev` is mentioned. `\b` (word boundary) matches between a word character and a hyphen, so `/@bot\b/` incorrectly matches `@bot-dev`. Replaced with `(?![\w-])` to exclude hyphens. ## Test plan <!-- How did you verify the changes? --> ## Checklist - [x] All commits are signed and verified - [x] `pnpm validate` passes - [x] Changeset added (or N/A — see [CONTRIBUTING.md](./CONTRIBUTING.md)) - [x] Documentation updated (or N/A) --------- Signed-off-by: sivchari <shibuuuu5@gmail.com> Co-authored-by: Ben Sabic <bensabic@users.noreply.github.com> |
||
|
josh
|
4bca64f058 |
feat(x): support image uploads on posts and DMs (#700)
## summary
the X adapter previously rejected every attachment (`File uploads are
not supported by the X adapter yet`). this adds image upload support:
images passed as `files` or `attachments` are uploaded through X's v2
chunked media endpoints and attached to the resulting post or DM
- uploads via X's current path-based flow: `POST
/2/media/upload/initialize` (JSON) then `/{id}/append` (multipart) then
`/{id}/finalize`, reusing the adapter's managed OAuth token
- attaches `media_ids` on `POST /2/tweets` for posts, and `attachments`
for DMs
- supports png, jpeg, and webp, up to 4 per post, with or without text
- requires the `media.write` OAuth 2.0 scope on the user token
### before / after
- before: posting a message with `files`/`attachments` throws a
`ValidationError`
- after: images upload and attach, and a post can be media-only or media
plus text
<details>
<summary>usage</summary>
```typescript
await thread.post({
markdown: "France lead the title race",
files: [{ data: pngBuffer, filename: "odds.png", mimeType: "image/png" }],
});
```
</details>
## test plan
- added unit tests covering the initialize JSON body, the multipart
append path, finalize, `media_ids` on the tweet, DM `attachments`,
media-only posts, MIME inference from filename, the over-limit
rejection, and unsupported-type rejection
- verified live against the X API end to end: uploaded an image and
posted then deleted it through the adapter (the initial command-param
implementation 400'd against the live API, which is what surfaced the
path-based endpoints as required)
- `pnpm --filter @chat-adapter/x build`, tests, `pnpm exec biome check`,
and `pnpm konsistent` all pass
---------
Signed-off-by: dancer <josh@afterima.ge>
|
||
|
ishaksebsib
|
8d7ccdb11b |
feat(telegram): support multiple file and attachment uploads (#605)
## Summary Adds Telegram media group support for posting multiple files and compatible typed attachments. Multiple `files` are sent as document media groups, while `attachments` preserve image, video, audio, or file media types and use Telegram’s native `sendMediaGroup`. --------- Signed-off-by: dancer <josh@afterima.ge> Co-authored-by: dancer <josh@afterima.ge> |
||
|
Dean Mauro
|
0fdb902980 |
feat(discord): Components support (#678)
## Summary #### What Adds opt-in support for Discord [Components](https://docs.discord.com/developers/components/reference). #### Why Discord Components allows developers more control over the layout of bot messages by treating text, images, files, and buttons as flexible components. Instead of the rigid text-above-embeds layout, elements can be arranged in any order or column. An extreme example of what's possible with Components: <img width="728" height="1117" alt="image" src="https://github.com/user-attachments/assets/fde6ab9c-f635-45fb-b64f-9ddf4c26580f" /> #### How Embeds remain the default behavior. The Discord adapter now supports a `componentsV2` flag to When enabled, card messages render with Discord Components v2 containers, sections, text displays, media galleries, separators, buttons, and string selects, and include the `IS_COMPONENTS_V2` message flag. ## Test plan Create a Chat with the Discord adapter. Set `contentFormat: DiscordContentFormat.ComponentsV2` and create a post that uses sections, markdown, buttons, etc. Note that 1. All elements should render correctly. 2. Individual sections can contain their own actions. 3. Markdown formatting gets rendered correctly. ``` import { Actions, Button, Card, CardText, Image, LinkButton, Section } from "chat"; import { createDiscordAdapter } from "@chat-adapter/discord"; const discord = createDiscordAdapter({ contentFormat: DiscordContentFormat.ComponentsV2, }); await thread.post( <Card title="Deployment ready" subtitle="Production build completed"> <Section> <CardText> **Version 2.4.0** is ready to promote. Review the release notes, then choose an action below. </CardText> <Image url="https://example.com/deploy-preview.png" alt="Preview" /> </Section> <Actions> <Button id="promote" style="primary"> Promote </Button> <Button id="rollback" style="danger"> Roll back </Button> <LinkButton url="https://example.com/deployments/123"> View deployment </LinkButton> </Actions> </Card> ); ``` --------- Signed-off-by: dancer <josh@afterima.ge> Co-authored-by: vercel[bot] <35613825+vercel[bot]@users.noreply.github.com> Co-authored-by: dancer <josh@afterima.ge> |
||
|
Max
|
6c2a3918d5 |
feat(discord): rename thread channels (#693)
## Summary Add `DiscordAdapter#setThreadTitle()` to rename native Discord thread channels and document the required **Manage Threads** permission. Signed-off-by: onmax <maximogarciamtnez@gmail.com> Signed-off-by: dancer <josh@afterima.ge> Co-authored-by: dancer <josh@afterima.ge> |
||
|
Faraz Patankar
|
5341f909a4 |
feat(discord): ignore @everyone/@here pings unless respondToGlobalMentions is set (#701)
## Summary Fixes #699. In gateway mode, the bot responded to `@everyone`/`@here` announcements. The legacy gateway listener used discord.js `message.mentions.has(botId)`, which counts global pings as a mention of the bot by default. Global pings are now ignored on both gateway paths by default, and a new `respondToGlobalMentions` config option (default `false`, per the issue's suggestion) lets users opt back in: - **Legacy gateway mode**: mention detection passes `{ ignoreEveryone: true }` to `mentions.has()`; `@everyone`/`@here` only count via `message.mentions.everyone` when the option is enabled. Direct mentions, `mentionRoleIds` role mentions, and replies to the bot are unaffected. - **Forwarded gateway mode**: added the `mention_everyone` field to `DiscordGatewayMessageData` and wired it into the same gate (this path previously ignored global pings silently, with no way to opt in). ## Changes - `respondToGlobalMentions?: boolean` on `DiscordAdapterConfig` (default `false`) - Mention detection updated in both `setupLegacyGatewayHandlers` and `handleForwardedMessage` - 5 new tests covering both transports: global pings ignored by default, honored when opted in, and direct mentions still detected while global pings are ignored - Docs: config tables in `apps/docs/content/adapters/official/discord.mdx` and the adapter README - Changeset (minor, `@chat-adapter/discord`) ## Testing - `pnpm validate` passes (knip + check + typecheck + test + build) - All 249 `@chat-adapter/discord` tests pass, including the 5 new ones 🤖 Generated with [Claude Code](https://claude.com/claude-code) Signed-off-by: Faraz Patankar <farazpatankar@gmail.com> Co-authored-by: Claude Fable 5 <noreply@anthropic.com> |
||
|
Ben Sabic
|
4717a38407 |
feat(slack): support data table and data visualization blocks (#696)
Adds support for Slack's [data table](https://docs.slack.dev/reference/block-kit/blocks/data-table-block) and [data visualization](https://docs.slack.dev/reference/block-kit/blocks/data-visualization-block) Block Kit blocks. - **`chat`**: new cross-platform `ChartElement` + `Chart()` builder (JSX supported) mirroring Slack's model — pie `segments`, or bar/area/line `series` against shared `categories`. `Table()` gains optional `caption` and `pageSize`. Charts degrade to a text table on other platforms via the shared card fallback (`chartElementToFallbackText`). - **`@chat-adapter/slack`**: card tables now render as paginated, sortable `data_table` blocks by default (header-only tables keep the plain `table` block; oversized tables still fall back to ASCII). Charts render as `data_visualization` blocks; charts violating Slack constraints — including the undocumented **max 2 charts per message** — fall back to a text rendering instead of an API rejection. Same treatment in the `@chat-adapter/slack/blocks` subpath. - **`postMessage`** now surfaces Slack's per-block validation messages on `invalid_blocks` errors (this is how the 2-chart limit was found). - Example app gets a **Show Charts** button and table pagination on **Show Table**; docs, feature matrices, and changeset updated. Verified live against Slack: data table pagination/sorting and both chart types render natively. <table> <tr> <th>Data Table</th> <th>Data Charts</th> </tr> <tr> <td><img width="979" height="896" alt="CleanShot 2026-07-12 at 23 28 32" src="https://github.com/user-attachments/assets/3307bd90-9322-452f-86fb-07d46446822d" /></td> <td><img width="955" height="879" alt="CleanShot 2026-07-12 at 23 29 02" src="https://github.com/user-attachments/assets/ddb31a1b-e3fd-457c-a2e6-bde4934afebe" /></td> </tr> </table> --------- Co-authored-by: Ben Sabic <bensabic@users.noreply.github.com> |
||
|
damianborowy-nexos
|
1721fa01e7 |
feat(slack): add Slack Agent messaging experience (agent_view) support (#684)
## Summary
Add support for Slack's Agent messaging experience (`agent_view`), the
2026 replacement for `assistant_view`.
## Core (`chat`)
- New `onAppContextChanged` event carrying the active-view context as a
normalized `AppContextEntity[]` (`channel` / `canvas` / `list` /
`message` / `unknown`) describing what the user is currently viewing.
- `AppHomeOpenedEvent` now carries:
- the same folded active-view context as optional `entities`
- the opened `tab` (`"home"` / `"messages"`), so handlers can
distinguish a Home-tab open from the DM-open signal under `agent_view`
## Slack adapter (`@chat-adapter/slack`)
- **`agentView` config flag.** Under `agent_view`:
- `app_home_opened` is the DM-open signal and fires regardless of tab
(branch on `event.tab` if you also publish a Home view)
- DM messages are threaded per Slack's new model — each user message is
a thread root (`thread_ts ?? ts`)
- conversation-scoped threads returned by `openDM()` keep working: when
that thread is subscribed, incoming top-level DM messages route to it,
so `onSubscribedMessage` and per-thread state behave the same as in
legacy mode
- **`app_context_changed` routing** with normalized entities. Malformed
payloads degrade gracefully: a missing `context` yields `entities: []`,
and entities with a null/malformed `value` normalize to `kind:
"unknown"` — never a webhook 500.
- **`getAppContext(message)`** helper to read the folded active-view
context off a DM message.
- **`setSuggestedPrompts`** accepts an optional thread reference
(`agent_view` lets prompts sit at the top of the agent conversation).
- **Env auth fallback now keys off auth fields**: `SLACK_BOT_TOKEN` /
`SLACK_CLIENT_ID` / `SLACK_CLIENT_SECRET` fallback is disabled only when
an auth-related field (`botToken`, `clientId`, `clientSecret`,
`installationProvider`) is passed explicitly, rather than by the
presence of any config object. This lets non-auth options compose with
env auth — e.g. `createSlackAdapter({ agentView: true })` picks up env
credentials — and matches the semantics documented in the adapter's
AGENTS.md. *(Behavior change for callers passing non-auth-only configs
while relying on env vars being ignored.)*
- Bumped `@slack/web-api` to `^7.18.0` (adds the optional `thread_ts`
typing for `setSuggestedPrompts`).
## Docs
- New "Agent messaging experience" section on the Slack adapter page
(config, manifest snippet, threading model, openDM bridge).
- "Handling active-view context" section in handling-events, plus
`tab`/`entities` rows on the app-home event table.
- Callout: under `agent_view`, bot replies are threaded per user
message, so `conversations.history` only returns the user's side of a DM
— build AI conversation history from transcripts instead of channel
history.
## Example app (`examples/nextjs-chat`)
- Plain `SLACK_BOT_TOKEN` adapter branch (previously Slack was only
wired via Vercel Connect).
- DM AI history built from transcripts instead of channel history (see
docs callout above); assistant turns persisted.
- The `dm me` trigger regex now matches mention text, which carries the
`@bot` prefix on Slack.
## Test plan
- `pnpm validate` and `pnpm konsistent` pass.
- Unit tests cover the new events, entity normalization (including
malformed payloads), `agent_view` DM threading, the openDM subscription
bridge, `tab` passthrough, `setSuggestedPrompts` thread handling, and
env-fallback behavior; an integration replay test exercises the full
webhook flow.
- Verified manually against a live `agent_view` workspace:
`onAppContextChanged` entities, folded context on `app_home_opened` and
DM messages, `tab` values for both tabs, per-message DM threading, the
openDM subscription bridge, and signed malformed-payload replays (all
return 200).
- Legacy regression pass with `agentView` off: conversation-scoped DM
threading, Home-tab-only `app_home_opened`, mention flow unchanged.
### Slack references
- Agent messaging experience:
https://docs.slack.dev/changelog/2026/06/30/agent-messages-tab/
- Active-view context:
https://docs.slack.dev/changelog/2026/07/02/app-context/
## Checklist
- [x] All commits are signed and verified
- [x] All commits are signed off for the DCO (`git commit -s`)
- [x] `pnpm validate` passes
- [x] Changeset added
- [x] Documentation updated
---------
Signed-off-by: Damian Borowy <301205838+damianborowy-nexos@users.noreply.github.com>
Signed-off-by: Ben Sabic <bensabic@users.noreply.github.com>
Co-authored-by: Ben Sabic <bensabic@users.noreply.github.com>
|
||
|
Rich Haines
|
1dff4515e2 |
refactor(docs): migrate chat-sdk.dev to @vercel/geistdocs (#686)
## Summary Migrates `apps/docs` from locally-copied geistdocs runtime code to the published [`@vercel/geistdocs`](https://www.npmjs.com/package/@vercel/geistdocs) package (1.8.2), following the official [migration guide](https://preview.geistdocs.com/docs/migration). Net **−8,400 lines**. ### Package-backed now - Docs page + layouts: `createDocsPage`, `GeistdocsDocsLayout`, `GeistdocsHomeLayout` (JSON-LD + sr-only markdown hints preserved via `renderTop`) - Navbar (OSS product switcher via `navbarOssProducts`), footer, provider, search dialog, page actions (edit source, feedback, copy page, Ask AI, open-in-chat, scroll top) - `/api/search` → `createSearchRoute`, `/api/chat` → `createChatRoute` (AI SDK v6; AI Gateway default, optional `GEISTDOCS_CHAT_PROXY_URL`) - `llms.mdx` → `createDocsMarkdownRoute`, `sitemap.md` → `createSitemapMarkdownRoute` (now includes an **Adapters** section) - **New**: `/agents.md` via `createAgentsRoute`, backed by a new `agent` readiness config - `proxy.ts` → `createProxy` with explicit `markdownRoutes` for `/docs` → `llms.mdx` and `/adapters` → `adapters.mdx` (adds AI-agent UA rewrites) - CSS: `@vercel/geistdocs/styles.css` + slim local overrides (shadcn tokens for remaining `components/ui`, body tint, prose inline code, `#nd-*` tweaks); code blocks now use the geist Shiki theme - Icons/logos from `@vercel/geistdocs/assets/*`; feedback via the package action (same geistdocs.com endpoint + `siteId`) ### Kept local by design - Curated `/llms.txt` index + `/llms-full.txt` corpus — the published `AGENTS.md`/SKILL.md artifacts and integration tests reference this exact contract - The adapters section (README fetching, OG images, JSON-LD, feature matrices, `adapters.mdx` markdown route) — now rendered inside the package docs layout - RSS and OG image routes (app-owned per the migration guide) - Skipped `/.well-known/mcp.json`: no MCP servers configured, and the proxy matcher must keep excluding `.well-known` for the served agent-skills files ### Cleanup - Deleted local copies: `components/geistdocs/*` chrome, `components/ai-elements/*`, chat hooks/persistence, feedback server actions, unused shadcn primitives, geistcn logo/icon fallbacks covered by package assets - Removed 13 now-unused deps (`ai@5`, `@ai-sdk/react@2`, `dexie`, `jotai`, `cmdk`, `vaul`, `mermaid`, `nanoid`, `react-player`, `use-stick-to-bottom`, `@orama/tokenizers`, `dexie-react-hooks`, `next-themes`) - Updated `docs-llms.test.ts` proxy assertions to the `createProxy` markdown-route shape ### Behavior changes to be aware of - Code blocks use the geist Shiki theme instead of GitHub light/dark - Ask AI history is no longer persisted in IndexedDB (package owns the panel) - Adapters sidebar uses the standard geistdocs tree rendering instead of the bespoke grouped sidebar - Per-page markdown output appends the standard geistdocs footer links (`/sitemap.md`, `/llms.txt`, `/agents.md`) ## Test plan - `pnpm validate` green (knip + check + typecheck + test + build) - Smoke-tested against `next build && next start`: `/`, `/docs`, `/adapters`, `/agents.md`, `/llms.txt`, `/llms-full.txt`, `/sitemap.md`, page-level `.md` URLs for both docs and adapters, `Accept: text/markdown` negotiation, search API, JSON-LD, sr-only markdown hints, edit-source URLs (`apps/docs/content/docs/{path}`), OSS navbar, page actions - Verified compiled CSS chunks contain the home grid, Shiki palette, and geist utilities (note: stale turbopack dev caches from before this change can serve incomplete CSS — `rm -rf apps/docs/.next` fixes it) ## Checklist - [x] All commits are signed and verified - [x] All commits are signed off for the DCO (`git commit -s`) - [x] `pnpm validate` passes - [x] Changeset added (or N/A — docs app + tests only, no package behavior change) - [x] Documentation updated (or N/A) --------- Signed-off-by: molebox <rich@vercel.com> |
||
|
josh
|
b3123815fa |
docs: use xai/grok-4.5 across docs, guides, and examples (#687)
## summary - swap chat model strings to `xai/grok-4.5` across the docs site, the shipped guides, and the example bots, so the docs lead with the latest model - apps/docs: ai overview, ai-sdk-tools, and streaming pages, the two landing-page code samples, and the live chat demo route (`app/api/chat/route.ts`, previously `openai/gpt-4.1-mini`) - packages/chat/resources/guides: the seven guides that use a chat model (slack connect, slack + ai sdk, liveblocks, vercel blob, github code review, daily digest, ai gateway) - examples: nextjs-chat and nuxt-chat bots - deliberately left non-chat model strings as they were, since grok-4.5 cannot fill those roles: the `openai/text-embedding-3-small` embedding model and the `openai/gpt-4o-mini` reranker - in the ai gateway guide, grok-4.5 is now the primary model but the fallback list stays cross-provider (`anthropic/claude-opus-4.8`, `google/gemini-3.1-pro-preview`) so the failover example still demonstrates real cross-provider fallback - also fixed a pre-existing prose/code mismatch in that guide, the fallback prose said `claude-opus-4.7` while the code listed `4.8`, now aligned to `4.8` --------- Signed-off-by: dancer <josh@afterima.ge> |
||
|
github-actions[bot]
|
b1940d2374 |
chore(release): version packages (#660)
This PR was opened by the [Changesets release](https://github.com/changesets/action) 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/github@4.33.0 ### Minor Changes - |
||
|
josh
|
ef2542c5fd |
feat(x): add X (Twitter) adapter (#682)
## summary
new `@chat-adapter/x` adapter for X (Twitter), built on the X API v2 and
the X Activity API. write bot logic once and reply to mentions, hold DM
conversations, post from the account, and like posts, like the other
Chat SDK adapters
what it supports:
- reply to public mentions (`post.mention.create`) and top-level posts
via `channel.post`
- send and receive direct messages (`dm.received` / `dm.sent`)
- edit and delete owned posts, delete own DM events
- likes as the only reaction (`emoji.heart` or `"like"`)
- buffered streaming: accumulates an LLM stream and posts once instead
of post+edit churn on a public timeline
- OAuth 2.0 user context with managed token refresh (rotating refresh
token persisted in the state adapter, optional AES-256-GCM encryption)
- webhook CRC and `x-twitter-webhooks-signature` verification
key design decisions:
- DMs are threaded by the other participant's user id (`x:dm:{userId}`)
because X DM webhooks carry no conversation id, only participants
- OAuth 2.0 only at runtime: DM send and read are verified to work on
OAuth 2.0 user tokens, so no OAuth 1.0a in the adapter (subscription and
webhook setup is one-time and handled in the X developer console)
- parsers were written against real captured payloads: mentions use the
v2 shape (author hydrated in `includes.users`), DMs use the legacy
Account Activity shape (`direct_message_events`,
`message_create.message_data`, a `users` map, and no conversation id)
also includes the `chat/adapters` catalog entry, docs page, CLI scaffold
spec, and `sample-messages.md` with real captured payloads
<details><summary>usage</summary>
```typescript
import { Chat } from "chat";
import { createXAdapter } from "@chat-adapter/x";
const bot = new Chat({
userName: "mybot",
adapters: { x: createXAdapter() },
});
bot.onNewMention(async (thread, message) => {
await thread.post(`hi @${message.author.userName}!`);
});
bot.onDirectMessage(async (thread) => {
await thread.post("hello from X");
});
```
</details>
## test plan
- adapter unit tests pass against the real captured payload shapes, with
regression tests for author-from-`includes` (mentions) and the legacy
`direct_message_events` shape (DMs)
- real captured `post.mention.create` and `dm.received` payloads
verified end-to-end through `handleWebhook`: signature verification,
routing, author resolution, and participant threading, plus
bad-signature rejection returns 401
- every write and read path fired live against the X API through the
adapter: top-level post, reply to a mention, like and unlike, edit,
delete, DM send, DM read, DM delete
- OAuth 2.0 managed token refresh exercised live (access and refresh
token rotation)
---------
Signed-off-by: dancer <josh@afterima.ge>
|
||
|
Mars Xiang
|
6de45723ef |
fix(discord): Implement rehydrateAttachment (#679)
## Summary The Discord adapter never implemented `rehydrateAttachment`, so consumers couldn't rebuild an attachment's `fetchData` after a message was serialized and restored, and inbound Discord attachments couldn't be downloaded once rehydrated. Every other URL/media adapter implements it. Discord attachment URLs are directly fetchable and survive serialization, so `fetchData` is rebuilt to fetch the URL (signed params preserved, no auth header — the links are pre-signed), reading `fetchMetadata?.url ?? attachment.url` like the other URL-based adapters. Returns the attachment unchanged when there's no URL. Adds unit tests and a changeset. ## Test plan - `pnpm --filter @chat-adapter/discord test` — 248 passing - `pnpm validate` Signed-off-by: marsxiang5902 <marsxiang5902@gmail.com> |
||
|
Ben Sabic
|
e7a396ae70 |
feat(tests): add threadId + self-message contracts and adopt across adapters (#675)
## @chat-adapter/tests Adds two shared behavioral contracts (self-tested against fakes, exported, with a minor changeset): - `threadIdContract` — verifies a thread-id codec round-trips (`decode(encode(x))`), prefixes ids with the adapter name, matches pinned encoded strings, and optionally distinguishes DM from non-DM. - `selfMessageContract` — verifies an adapter dispatches inbound messages from other users but ignores messages the bot authored itself (uses the shared matchers). ## Adoption - `threadIdContract` adopted across 10 adapters (github, linear, gchat, teams, discord, slack, telegram, whatsapp, twilio, messenger), replacing bespoke encode/decode/round-trip/isDM blocks while keeping error/edge cases. - `selfMessageContract` adopted where it cleanly applies (github, messenger). Deliberately skipped where adapters dispatch-then-flag `isMe` (linear, slack, gchat) or lack a network-free webhook self test (teams, whatsapp, etc.). Net ~−290 more lines. Stacked on #674. --------- Signed-off-by: Ben Sabic <bensabic@users.noreply.github.com> Co-authored-by: Ben Sabic <bensabic@users.noreply.github.com> |
||
|
Ben Sabic
|
ac5a54ee1d |
test(adapters): adopt shared @chat-adapter/tests factories and matchers (#674)
Wires `@chat-adapter/tests` as a devDependency and registers its matchers via `setupFiles: ["@chat-adapter/tests/setup"]` across all 11 platform adapters, then replaces bespoke local `mockLogger`/`createMockState`/`createMockChatInstance` with the shared factories and adopts `toHaveDispatched`/`not.toHaveDispatched` where clean. - 10 adapters migrated (gchat, messenger, teams, whatsapp, telegram, discord, twilio, linear, github, slack). Positional `createMockChatInstance(...)` call sites converted to the options API (slack 100, linear 35). - `web` left as-is — its suite uses the real `Chat`/`createMemoryState` for e2e, so the shared factories don't apply. - Platform SDK mocks (Octokit, WebClient, socket-mode, `@linear/sdk`, `fetch`) and the Phase 1 `connectWebhookContract` descriptors are left intact. Net ~−540 lines of duplicated test scaffolding. Stacked on #673. Tests-only, no changeset. --------- Signed-off-by: Ben Sabic <bensabic@users.noreply.github.com> Co-authored-by: Ben Sabic <bensabic@users.noreply.github.com> |
||
|
Ben Sabic
|
840c0d16e9 |
test(adapters): migrate Vercel Connect webhook tests to connectWebhookContract (#673)
Adopts the shared `connectWebhookContract` from `@chat-adapter/tests` in the Slack, GitHub, and Linear suites, replacing the bespoke `webhookVerifier` blocks (verifier pass → 200, throw/falsy → 401, invoked with request + raw body, precedence over a native secret). Adapter-specific Connect tests are kept (token resolvers, GitHub bot-id capture, type-level mutual exclusivity, 400-on-invalid-JSON, Linear identity/`withInstallation`). Each descriptor keeps `initialize()` network-free (GitHub `botUserId`, Slack `_botUserId` to skip `auth.test`, Linear stubs `resolveConnectIdentity`). Twilio is intentionally not included — it has a single generic `webhookVerifier` usage with no 200/401 gating suite to migrate. First of three stacked test-generalization PRs. Tests-only, no changeset. Signed-off-by: Ben Sabic <bensabic@users.noreply.github.com> Co-authored-by: Ben Sabic <bensabic@users.noreply.github.com> |
||
|
zaadevofc
|
eb466e526f |
docs: add chat-adapter-zaileys community adapter (#677)
## Summary Adds **chat-adapter-zaileys** to the community adapters catalog — a WhatsApp adapter powered by [Zaileys](https://github.com/zeative/zaileys), a batteries-included TypeScript wrapper around the unofficial WhatsApp Web API. - npm: https://www.npmjs.com/package/chat-adapter-zaileys - Repo: https://github.com/zeative/chat-adapter-zaileys - Docs: https://zeative.github.io/chat-adapter-zaileys/ ## What it adds vs the existing Baileys community adapter - Real `thread.fetchMessages` history backed by a pluggable message store (memory/SQLite/Postgres/Redis/Convex), with cursor pagination and `rehydrateAttachment` for queue/debounce strategies - Cards render as **native WhatsApp buttons** with `chat.onAction` round-trips - Poll votes decrypted natively — no `messageSecret` bookkeeping, works across restarts - `scheduleMessage` support (persisted scheduler) - Opt-in slash-command routing to `chat.onSlashCommand` - QR/pairing auth, reconnection, and session persistence handled by the underlying client ## Files changed (per `.agents/skills/add-adapter`) - `apps/docs/content/adapters/community/zaileys.mdx` — docs page with feature matrix - `apps/docs/content/adapters/community/meta.json` — slug added to Platforms - `apps/docs/adapters.json` — registry entry - `packages/integration-tests/src/documentation-test-utils.ts` — `chat-adapter-zaileys` + `zaileys` in `VALID_DOC_PACKAGES` ## Validation - `pnpm --filter chat build` ✓ - `pnpm --filter @chat-adapter/integration-tests test` → 914/914 ✓ - `pnpm --filter chat typecheck` ✓ - `pnpm check` + `pnpm konsistent` ✓ Signed-off-by: zeative <zaadevofc@gmail.com> |
||
|
OmriBenShoham
|
0c761f1bdd |
docs(adapters): add Dial as vendor-official adapter (#676)
Adds Dial as a vendor-official adapter — SMS, MMS, iMessage, and inbound voice-call transcripts for Chat SDK. - `vendor-official/dial.mdx` adapter page (following the Photon / Linq / Sendblue format) - catalog entry in `packages/chat/src/adapters/index.ts` with `DIAL_API_KEY` / `DIAL_FROM_NUMBER_ID` / `DIAL_WEBHOOK_SECRET` - `create-chat-sdk` scaffold spec entry - registry entry in `adapters.json` + `dial` added to vendor-official `meta.json` - integration-test doc lists + changeset Repo: https://github.com/GetDial-AI/chat-sdk-adapter · npm: `@getdial/chat-sdk-adapter` · Dial docs: https://docs.getdial.ai/integrations/agent-clients/vercel-chat-sdk The adapter maps a phone conversation to a Chat SDK thread (identified by the pair of phone numbers — Dial-owned and peer), an SMS/MMS/iMessage to a message with optional media attachments, and a completed voice call's transcript to a message on the caller's thread. Outbound sends and transcript fetches go through the official `@getdial/sdk`; inbound webhooks are HMAC-SHA256 verified against a per-subscription signing secret with constant-time compare. ### Validation - `pnpm --filter chat build` — clean - `pnpm --filter chat typecheck` — clean - `pnpm --filter create-chat-sdk typecheck` — clean - `pnpm --filter @chat-adapter/integration-tests exec vitest run src/docs-adapters.test.ts` — 361/361 passed - `pnpm check` (ultracite) — clean - `pnpm konsistent` — 34 files, no violations |
||
|
Ben Sabic
|
d4c52cade3 |
refactor(shared): share the bare-mention scanner across Discord, Teams, and Slack (#652)
The Discord adapter (fixed in #651) converts bare `@mentions` with a regex. A single-character lookbehind can't tell whether an `@` sits inside a URL, a code span, or an email host, so it still mangles cases the regex can't see. The Slack adapter already had a robust character-scanning resolver — `replaceBareMentions` — that handles exactly those cases, but it lived inside `adapter-slack`. This lifts that scanner into `@chat-adapter/shared` and points every adapter that does bare-mention conversion at it: Slack (dedup, no behavior change), Discord, and Teams — which had the identical `/@(\w+)/g` → `<at>$1</at>` bug in two places. ## What changed - **`@chat-adapter/shared`** — new `replaceBareMentions` (+ `MentionReplacer` type), moved verbatim from Slack. It skips inline/fenced code, scheme + schemeless URLs, and existing `<…>` tokens before handing each real `@name` to a platform-specific replacer. Adds a dedicated test file (the scanner had no direct tests before). - **Slack** — sources `replaceBareMentions` from `@chat-adapter/shared`; local `mentions.ts` deleted. Behavior unchanged. - **Discord** — regex → scanner in both conversion sites. - **Teams** — same fix for `<at>…</at>` mention tags. ## What this fixes (Discord + Teams) Across both the `{markdown}`/AST and `{raw}`/plain-string paths: | input | before | after | |---|---|---| | `https://github.com/@vercel` | `https://github.com/<@vercel>` | preserved | | `twitter.com/@jack` | `twitter.com/<@jack>` | preserved | | `` `ping @here` `` | `` `ping <@here>` `` | preserved | | `<@123>` (Discord, raw) | `<<@123>>` | preserved | Emails (`user@example.com`), period-prefixed mentions (`docs.@everyone`), and existing tokens keep working. ## Notes - `.changeset/config.json` uses `fixed: [["chat", "@chat-adapter/*"]]`, so the `@chat-adapter/shared` **minor** bump carries the whole family to a minor release; the `discord` / `teams` `patch` changesets exist for their changelog text. Slack has no behavioral change, so it gets no changeset. - Rebased on top of #651. That PR's `discord-email-mentions.md` changeset stays; this PR's scanner supersedes its regex implementation, so the two Discord changelog entries read as a progression. Signed-off-by: Ben Sabic <bensabic@users.noreply.github.com> Co-authored-by: Ben Sabic <bensabic@users.noreply.github.com> |
||
|
Thomas Dietert
|
0b63791b66 |
fix(slack): process Socket Mode retry envelopes instead of dropping them (#667)
Fixes #666 ## Summary Both `slack_event` handlers (`startSocketMode` and `runSocketModeListener`) ack and discard every envelope with `retry_num > 0`. Slack retries an event (immediately, +1 min, +5 min) when a prior delivery wasn't acked — including events that arrived while the app had **no open socket** (restart, deploy, or Slack's routine connection refreshes). For those, the retry is the only delivery the app ever sees, so dropping it permanently loses the event (production incident details in #666). - **`@chat-adapter/slack`**: route retry envelopes through `routeSocketEvent` like first deliveries (it acks per envelope type, preserving the 3s ack window), and log them at info with `retry_num` / `retry_reason` so redelivery is observable. Duplicate protection is unchanged and sufficient: `Chat.processMessage` dedupes on `message.id` (the Slack event `ts`, identical on a retry) via `state.setIfNotExists`. - **`chat`**: raise the default `DEDUPE_TTL_MS` from 5 to 10 minutes. Slack's final retry fires ~5 minutes after the original delivery — exactly at the old TTL boundary, where the dedupe entry from the first processing could expire just before the retry arrives and cause a double-process. `dedupeTtlMs` config still overrides. Behavior note for review: apps that relied on retries being invisible will now see redelivered events flow through — deduped when already handled, processed when not. That is the intended semantic: at-least-once delivery from Slack, exactly-once handling via the SDK's dedupe. ## Test plan - Replaced the `"skips retries"` test with `"processes retries like first deliveries (dedupe drops true duplicates)"` — asserts a `retry_num: 1` envelope is acked and reaches `processMessage`. - Updated the default-TTL test to 10 minutes; the custom-`dedupeTtlMs` test is unchanged. - `pnpm validate` passes end to end (knip, check, typecheck, test, build); `pnpm --filter chat --filter @chat-adapter/slack test` = 1028 + 506 passing. Signed-off-by: tdietert <thomasd@mercury.com> |
||
|
Ben Sabic
|
3abdc69103 |
docs(adapters): add Cloudflare Agents as vendor-official state adapter (#669)
Adds Cloudflare Agents as a vendor-official **state** adapter — `agents/chat-sdk`'s `createChatSdkState()`, a Chat SDK `StateAdapter` that stores subscriptions, locks, queues, dedupe keys, thread/channel state, transcripts, and history in Durable Object SQLite via `ChatSdkStateAgent` sub-agents. - `vendor-official/cloudflare-agents.mdx` state-adapter page (Agent setup, wrangler DO migration, sharding, config, storage/cleanup) - catalog entry in `packages/chat/src/adapters/index.ts` (`group: vendor-official`, `type: state`) - registry entry in `adapters.json` + `cloudflare-agents` in vendor-official `meta.json` - integration-test doc lists + changeset Repo: https://github.com/cloudflare/agents · package `agents` (`agents/chat-sdk`) · [docs](https://developers.cloudflare.com/agents/runtime/communication/chat-sdk/) ### Not wired into the create-chat-sdk CLI This adapter runs inside a Cloudflare Worker with Durable Objects, not the generated Next.js runtime, so it is intentionally kept out of the scaffold: - added to `CLI_INCOMPATIBLE_ADAPTERS` (rejected via `--adapter`, hidden from the platform picker and e2e run, like `lark`/`matrix`) - new `listCliStateAdapters()` filters the interactive **state** picker and the `--help` adapter list (the state picker previously used raw `listStateAdapters()` and would have offered it, then thrown on selection) ### Tests - `catalog/display.test.ts` — `listCliStateAdapters`: returns only state adapters, includes `memory`/`redis`, and excludes `cloudflare-agents` while asserting it *is* in the raw catalog - `catalog/selection.test.ts` — `resolveAdapterValue("cloudflare-agents")` throws "not supported" - `cli/program.test.ts` — `buildAdapterList()` help text omits `cloudflare-agents` - existing `CLI_SCAFFOLD_SPEC covers every catalog adapter` + docs-adapters/docs-content suites cover the catalog entry, registry parity, and MDX imports ### Validation - create-chat-sdk: **178 passed**, typecheck clean - integration docs suites pass; Biome + knip clean --------- Co-authored-by: Ben Sabic <bensabic@users.noreply.github.com> |
||
|
Ben Sabic
|
24a04d5653 |
docs(adapters): add Photon as vendor-official adapter (#668)
Adds Photon as a vendor-official adapter — iMessage for Chat SDK. - `vendor-official/photon.mdx` adapter page (following the Linq / Sendblue / Kapso format) - catalog entry in `packages/chat/src/adapters/index.ts` with cloud/self-host credential modes - `create-chat-sdk` scaffold spec entry - registry entry in `adapters.json` + `photon` added to vendor-official `meta.json` - integration-test doc lists + changeset Repo: https://github.com/photon-hq/vercel-chat-adapter-imessage · npm: `@photon-ai/chat-adapter-imessage` · built on [spectrum-ts](https://github.com/photon-hq/spectrum-ts) The adapter runs in three modes — **Cloud** ([Spectrum Cloud](https://app.photon.codes)), **self-hosted** (gRPC), and **local** (on-device, macOS) — auto-detected from environment variables. Cloud mode delivers inbound messages via HMAC-signed webhooks; DMs can be replied to cold from a webhook delivery. ### Notes - Catalog slug is `photon`; docs code examples use `imessage` as the adapter key to match the upstream README. - Feature flags encode the README's remote-only caveats (reactions / editing / typing / modals as `partial`, mentions as DMs-only; no history, thread info, or reaction removal). ### Validation - `docs-adapters` integration tests — 1237 passed (catalog↔registry parity, peerDeps↔PackageInstall alignment) - `create-chat-sdk` e2e scaffold — 175 passed (scaffolds every catalog adapter, incl. photon) - `chat` + `create-chat-sdk` typecheck, Biome check, and konsistent — clean Co-authored-by: Ben Sabic <bensabic@users.noreply.github.com> |
||
|
Ben Sabic
|
ba375ce16c |
feat(create-chat-sdk): add Vercel Connect mode (#655)
Adds an opt-in Vercel Connect authentication mode to the scaffolder for the Slack, GitHub, and Linear adapters, via a `--connect` flag and a new interactive auth-mode prompt (shown only when a Connect-capable adapter is selected). When enabled, the generated project: - spreads the matching helper from `@vercel/connect/chat` into the adapter factory in `src/lib/bot.ts` (non-Connect adapters keep their native factory calls) - adds `@vercel/connect` to dependencies - lists each connector UID (for example `SLACK_CONNECTOR`) plus the recommended `GITHUB_BOT_USER_ID`, in place of native provider secrets, in `.env.example` - documents `vercel link` / `vercel env pull` and the deployed-URL webhook caveat in the README and post-install next steps Connect policy lives in the existing `scaffold-spec.ts` (per-adapter `connect` field), so `chat/adapters` stays the single source of adapter metadata. Stacked on #647 (base `vercel-connect/base`). ## Companion `@vercel/connect/chat` subpath: vercel/vercel#16826. --------- Signed-off-by: Ben Sabic <bensabic@users.noreply.github.com> Co-authored-by: Ben Sabic <bensabic@users.noreply.github.com> |
||
|
Ben Sabic
|
a7fb1bcfa1 |
feat(tests): add Vercel Connect webhook contract helper (#654)
Adds `connectWebhookContract` to `@chat-adapter/tests`: a shared Vitest suite that verifies a Connect-capable adapter's webhook verification. Given a small per-adapter descriptor, it asserts the behavior every Connect adapter shares — a `webhookVerifier` replaces the native signature/secret check and gates inbound requests (`200` on a truthy result, `401` on a thrown error or falsy result) and is invoked with the request and raw body. The helper depends only on `chat` types and is self-tested against a fake adapter, so it lives entirely under `packages/tests`. Stacked on #647 (base `vercel-connect/base`). ## Follow-up The Slack, GitHub, and Linear adapters don't consume this helper yet — they still have their own inline `webhookVerifier` tests. Migrating those to `connectWebhookContract` will happen as part of the ongoing effort to port adapter tests over to `@chat-adapter/tests`, which is where the helper becomes real shared coverage. ## Companion `@vercel/connect/chat` subpath: vercel/vercel#16826. --------- Signed-off-by: Ben Sabic <bensabic@users.noreply.github.com> Co-authored-by: Ben Sabic <bensabic@users.noreply.github.com> |
||
|
Ben Sabic
|
6750d59e72 |
feat(github): add Vercel Connect support (#650)
Adds Vercel Connect support to the GitHub adapter: - A new `installationToken` config option (string or resolver) supplies installation access tokens directly, skipping the GitHub App private-key JWT exchange. - A new optional `webhookVerifier` verifies inbound webhooks (Connect trigger-forwarded requests via a Vercel OIDC token) in place of the GitHub webhook secret. Pair with `connectGitHubAdapter()` from `@vercel/connect/chat`. Includes a changeset (`@chat-adapter/github` minor). Stacked on #647 (base `vercel-connect/base`). ## Companion `@vercel/connect/chat` subpath: vercel/vercel#16826. <img width="933" height="755" alt="CleanShot 2026-06-30 at 12 02 18" src="https://github.com/user-attachments/assets/cc834560-0486-4f09-b8d5-8264be360544" /> --------- Signed-off-by: Ben Sabic <bensabic@users.noreply.github.com> Co-authored-by: Ben Sabic <bensabic@users.noreply.github.com> |
||
|
Ben Sabic
|
4115c9431e |
feat(linear): add Vercel Connect support (#649)
Adds Vercel Connect support to the Linear adapter: - `accessToken` now accepts a `() => string | Promise<string>` resolver in addition to a string, so tokens can be sourced from Vercel Connect at runtime. - A new optional `webhookVerifier` verifies inbound webhooks (Connect trigger-forwarded requests via a Vercel OIDC token) in place of the Linear webhook secret. - Connect-mode outbound calls outside webhook handling are supported via `withInstallation(organizationId, fn)`. Pair with `connectLinearAdapter()` from `@vercel/connect/chat`. Includes a changeset (`@chat-adapter/linear` minor). Stacked on #647 (base `vercel-connect/base`). ## Companion `@vercel/connect/chat` subpath: vercel/vercel#16826. <img width="929" height="664" alt="CleanShot 2026-06-30 at 12 35 30" src="https://github.com/user-attachments/assets/c5861cb9-d66b-42c6-b838-5b4983f48646" /> --------- Signed-off-by: Ben Sabic <bensabic@users.noreply.github.com> Co-authored-by: Ben Sabic <bensabic@users.noreply.github.com> |
||
|
Ben Sabic
|
ba687cb13c |
docs(slack): document Vercel Connect support (#648)
Documents authenticating the Slack adapter with Vercel Connect via `connectSlackAdapter()` from `@vercel/connect/chat`. The Slack adapter already supports a `botToken` resolver and a `webhookVerifier`, so this is a documentation-only change (no changeset). Stacked on #647 (base `vercel-connect/base`). ## Companion `@vercel/connect/chat` subpath: vercel/vercel#16826. <img width="824" height="527" alt="CleanShot 2026-06-30 at 12 03 26" src="https://github.com/user-attachments/assets/cbced069-8913-4848-9cf1-df0e5f614353" /> --------- Signed-off-by: Ben Sabic <bensabic@users.noreply.github.com> Co-authored-by: Ben Sabic <bensabic@users.noreply.github.com> |
||
|
Ben Sabic
|
ab0e1806c8 |
feat(chat): Vercel Connect (#647)
Adds a Vercel Connect guide to the docs under **Usage** (`chat-sdk.dev/docs/vercel-connect`), covering connector setup, trigger forwarding, the per-platform `connect*Adapter` helpers from `@vercel/connect/chat`, custom OIDC webhook verification and its trust boundary, and limitations. Also adds a "Vercel Connect Guide" card to the docs homepage. This is the base of a stack; the adapter, tests, and CLI PRs below build on it. Docs-only, so no changeset. ## Stack - #647 — feat(chat): Vercel Connect (this PR, base → `main`) - #648 — docs(slack): document Vercel Connect support - #649 — feat(linear): add Vercel Connect support - #650 — feat(github): add Vercel Connect support - #654 — feat(tests): add Vercel Connect webhook contract helper - #655 — feat(create-chat-sdk): add Vercel Connect mode All of the above are stacked on this branch (`vercel-connect/base`). ## Companion The helpers this documents ship in the `@vercel/connect/chat` subpath: vercel/vercel#16826. --------- Signed-off-by: Ben Sabic <bensabic@users.noreply.github.com> Co-authored-by: Ben Sabic <bensabic@users.noreply.github.com> |
||
|
josh
|
076fe5dc43 |
fix(chat): preserve skipped mention routing (#659)
## summary fixes skipped mention routing for collapsed concurrency messages queue and burst already passed skipped message context, but mention routing could still swallow message pattern handlers when no `onNewMention` handler was registered this also makes debounce preserve skipped context so an earlier debounced bot mention can still route to `onNewMention` when the latest message does not mention the bot |
||
|
github-actions[bot]
|
6f18930cf3 |
chore(release): version packages (#623)
This PR was opened by the [Changesets release](https://github.com/changesets/action) 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/discord@4.32.0 ### Minor Changes - |
||
|
josh
|
219625724c |
fix(discord): sort imports (#657)
## summary sorts the Discord adapter type imports so main lint passes again ## test plan - `pnpm exec ultracite check packages/adapter-discord/src/index.ts` - `git diff --check` |
||
|
Anudeep
|
2e4735118e |
fix: let Plan tasks run in parallel without implicit auto-completion (#632)
## Summary Plan’s task list API always marked existing in-progress steps as complete whenever a new step was added. That made sense for simple sequential bots, but it blocked parallel work — even though the docs already showed a parallel pattern and per-task updates by ID were added earlier. This PR adds an optional flag on task creation so callers can keep multiple steps in progress at once, while leaving the old sequential behavior as the default. **Opt-out flag, default on**. We considered removing auto-completion entirely. That would’ve been cleaner for parallel use but would’ve broken existing sequential bots that rely on implicit “move to next step” behavior. Defaulting to the current behavior keeps upgrades safe; parallel callers pass the flag off. **No broader API redesign**. Task completion stays explicit via status updates and the existing “complete plan” flow. The change is scoped to when a new task is appended. closes #630 |
||
|
Dean Mauro
|
022a502726 |
feat(discord): add ephemeral slash command responses (#514)
## summary resolves #515 adds Discord slash-command interaction response flags so selected commands can defer as ephemeral Discord locks ephemerality on the initial `DEFERRED_CHANNEL_MESSAGE_WITH_SOURCE` response, so the adapter now exposes `interactionFlags` on `createDiscordAdapter` for that initial acknowledgement ```ts import { createDiscordAdapter, DiscordInteractionResponseFlag, } from "@chat-adapter/discord"; const discord = createDiscordAdapter({ interactionFlags: ({ command }) => { if (command === "/admin") { return DiscordInteractionResponseFlag.Ephemeral; } }, }); ``` handlers still use the normal `event.channel.post(...)` flow, and `event.channel.postEphemeral(...)` keeps the normal Chat SDK fallback behavior outside Discord's slash-command interaction response path Co-authored-by: dancer <josh@afterima.ge> |
||
|
josh
|
eccc6b91bf |
fix(chat): detect mentions in skipped queued messages (#656)
## summary fixes #613 detects bot mentions across queued and burst skipped messages before routing handlers this makes `onNewMention` fire when an earlier skipped message mentions the bot and the latest collapsed message does not, while preserving `message.isMention` on the latest message adds regression coverage for both `queue` and `burst` |
||
|
Max
|
4ee187ac3c |
fix(telegram): start typing before message processing (#612)
## Summary Send Telegram typing actions immediately for private incoming message and slash command updates before handing the message to Chat SDK processing. This lets Telegram clients show the native `...` indicator during early processing instead of waiting for downstream handler code to call `thread.startTyping()`. The change is scoped to private, non-bot Telegram messages and adds regression coverage for normal messages and slash commands plus a patch changeset. Closes #611 ## Test plan - [x] `pnpm validate` - [x] `pnpm --filter @chat-adapter/telegram test` - [x] `pnpm --filter @chat-adapter/telegram typecheck` - [x] `pnpm --filter @chat-adapter/telegram build` ## Checklist - [x] All commits are signed and verified - [x] `pnpm validate` passes - [x] Changeset added (or N/A — see [CONTRIBUTING.md](./CONTRIBUTING.md)) - [x] Documentation updated (or N/A) --------- Signed-off-by: dancer <josh@afterima.ge> Co-authored-by: dancer <josh@afterima.ge> |
||
|
Hiroki Osame
|
fc7df9c4cf |
fix(github): remove raw webhook payload logging (#500)
- Remove raw GitHub webhook body previews from adapter debug/error logs - Prevents webhook payload content from being copied into application logs when debug logging is enabled - No webhook routing or response behavior change; only log fields change ## Context The GitHub adapter logged a preview of incoming webhook request bodies while handling webhooks. Raw webhook payloads can contain repository metadata, user-authored issue or pull request text, URLs, installation details, and other provider-controlled content. Even at debug level, SDK logging should avoid copying raw provider payloads into application logs by default. ## Problem Debug logging should provide useful operational context without changing the privacy boundary of webhook data. The previous log emitted a raw body preview before signature verification. That meant an application with debug logging enabled could record payload content from both valid GitHub webhook events and invalid requests that were later rejected. This is unnecessary for normal webhook troubleshooting. Derived request metadata is enough to understand routing and parsing failures without retaining payload text. ## Changes The GitHub adapter no longer logs raw webhook bodies or body previews. Webhook logs now use bounded request-shape metadata: - `bodyBytes` - `contentType` - `eventType` - `signaturePresent` - `jsonParseStatus` for invalid JSON The change preserves signature verification, ping handling, JSON parsing, and event routing behavior. Regression tests cover invalid signature, invalid JSON, and valid webhook paths with token-shaped and customer-slug sentinel strings in the payload. The tests assert those sentinels, the full raw body, the old raw-body log message, and `bodyPreview` do not appear in logger calls. A patch changeset is included for `@chat-adapter/github`. ## Verification - `pnpm turbo build --filter @chat-adapter/github` - `pnpm --filter @chat-adapter/github test` - `pnpm --filter @chat-adapter/github typecheck` - `pnpm check` - `git diff --check` |
||
|
furgin
|
0d4e3ee490 |
fix(discord): render bare URLs as bare links, not masked links. (#567)
## Summary Discord only renders masked links `[text](url)` inside embeds. In a normal message, a bare URL converted to `[url](url)` shows up as literal text rather than a clickable link. In nodeToDiscordMarkdown's link branch, return the bare URL when the link's label equals its target (the bare-URL / autolink case); labeled links are unchanged. Adds regression tests for both bare URLs and <autolinks>. Fixes #565. |
||
|
Syed Osama Ali Shah
|
490fa00e87 |
fix(discord): don't mangle email addresses into mentions (#651)
## Bug The Discord adapter converts `@mentions` with `/@(\w+)/g` in two places — `DiscordFormatConverter.convertMentionsToDiscord` (plain/`raw` messages) and the text-node branch of `nodeToDiscordMarkdown` (markdown/AST messages): ```ts text.replace(/@(\w+)/g, "<@$1>"); ``` That pattern matches `@word` even when the `@` is preceded by a word character, so it rewrites **email addresses** and `word@word` handles into broken mentions: | input | before | after | |---|---|---| | `Contact me at user@example.com` | `Contact me at user<@example>.com` | `Contact me at user@example.com` | | `ping support@vercel.com` | `ping support<@vercel>.com` | `ping support@vercel.com` | | `hey @alice` | `hey <@alice>` | `hey <@alice>` (unchanged) | The Slack adapter already guards against exactly this with a word-boundary check (`replaceBareMentions`); the Discord converter didn't. ## Fix Introduce a shared top-level `BARE_MENTION_PATTERN = /(?<![\w@.])@(\w+)/g` (per AGENTS.md, regex literals live at top level) with a negative lookbehind, so only an `@` at a word boundary becomes a mention. Emails/handles are left intact; real bare mentions still convert. Used in both conversion sites. ## Test Adds a regression test in `markdown.test.ts` asserting `Contact me at user@example.com` round-trips through `toAst`/`fromAst` without becoming a mention. Includes a changeset (`@chat-adapter/discord` patch). Commit is signed (Verified) and DCO signed-off. --------- Signed-off-by: Osamaali313 <86572800+Osamaali313@users.noreply.github.com> Signed-off-by: Ben Sabic <bensabic@users.noreply.github.com> Co-authored-by: Ben Sabic <bensabic@users.noreply.github.com> |
||
|
Ben Sabic
|
99c598505f |
docs: refresh agent docs, README badges, and Chat SDK skill (#646)
- Replace npm version/download badges with Agent Stack and MIT badges on the root README and all published package READMEs - Streamline root `AGENTS.md`: fix title, add an accurate monorepo map, trim duplicated CONTRIBUTING/Ultracite/env-var content, and link to package-level `AGENTS.md` files - Slim the Chat SDK agent skill (`skills/chat/SKILL.md` and published copies) to defer to bundled docs, chat-sdk.dev, Vercel KB, and `llms.txt` instead of inlining CLI flags, quick-start code, and API tables - Polish root README copy (install examples, adapter/build links, Vercel Plugin URL, Vercel KB link, “Made by Vercel” badge) - Minor `CONTRIBUTING.md` fixes: simplify DCO wording, correct preview-branch proxy file references (`proxy.ts` vs middleware) --------- Signed-off-by: Ben Sabic <bensabic@users.noreply.github.com> Co-authored-by: Ben Sabic <bensabic@users.noreply.github.com> |
||
|
wong2
|
ef3f0f63bd |
docs: add Weixin community adapter (#638)
Adds the [`chat-adapter-weixin`](https://github.com/wong2/weixin-chat-adapter) community adapter (Weixin / WeChat iLink bot) to the docs. ### What's included - `apps/docs/content/adapters/community/weixin.mdx` — hand-authored adapter page following the existing community-adapter structure (install, quick start, long-polling note, QR login, env vars, config `TypeTable`, thread-ID format, capabilities/limitations, and `<FeatureSupport />`). - `apps/docs/adapters.json` — registry entry (`community: true`, author, pinned README commit). - `apps/docs/content/adapters/community/meta.json` — sidebar link under **Platforms**. ### Notes The adapter talks to Weixin's iLink bot HTTP JSON APIs directly. It uses long polling for inbound messages (no webhook) and requires a Chat SDK `StateAdapter` for cursor / context-token / dedupe / history. It's 1:1 only, so messages route through `onDirectMessage`. ### Verification `docs-adapters` (322) and `docs-llms` (129) integration tests pass. --------- Co-authored-by: Ben Sabic <bensabic@users.noreply.github.com> |
||
|
Akazawa Daisuke
|
438f5513b0 |
fix: avoid dummy message context for lightweight threads (#633)
## summary fixes #631 removes dummy `Message` casts from lightweight thread, action, and reaction paths when no incoming message context exists this keeps the existing `currentMessage` guard meaningful and prevents streaming through `chat.thread(threadId)`, `chat.openDM(...)`, action threads, and reaction threads from reading fields from an empty object when Slack lacks the thread or recipient context required by `chat.startStream`, the adapter now returns `null` before consuming the stream so Chat SDK can transparently use its post-and-edit fallback native Slack streaming remains available for webhook-created threads and DM threads with valid native stream context --------- Co-authored-by: dancer <josh@afterima.ge> |
||
|
Ben Sabic
|
64b66864b1 |
chore(changesets): ignore all example-* packages and enforce the convention (#626)
Replace the explicit per-example entries in the changesets `ignore` list with an `example-*` name glob (matched by micromatch). All example apps are private and never published, so listing them individually only adds version and changelog churn to release PRs, and each new example required editing this CODEOWNERS-gated file. Add an integration test that resolves the changesets config against the workspace and asserts every examples/* package is in the resolved ignore list and follows the `example-*` naming convention, so an off-convention example app fails CI instead of silently leaking into releases. --------- Co-authored-by: Ben Sabic <bensabic@users.noreply.github.com> |
||
|
Himanshu Garg
|
937cac989e |
fix(telegram): prevent truncation of valid URLs with entity-marker characters in MarkdownV2 links (#610)
## Summary Fixes `@chat-adapter/telegram` silently truncating valid MarkdownV2 messages whose link URLs contain an odd number of entity-marker characters (`_`, `*`, `~`). `trimToMarkdownV2SafeBoundary` counted unescaped markers anywhere outside code spans and treated an odd total as an unterminated entity. But per the [MarkdownV2 spec](https://core.telegram.org/bots/api#markdownv2-style), only `)` and `\` are special inside the `(...)` URL part of an inline link — so a message ending with e.g. `[Read more](https://example.com/page?utm_campaign=a&utm_source=b&utm_channel=c)` (3 raw underscores in the URL) was sliced mid-URL, rejected by Telegram with `can't parse entities: Can't find end of a URL`, and degraded to plain text by the markdown fallback — links and formatting silently stripped. Changes in `findUnescapedPositionsOutsideCode`: - Tracks a link-URL state alongside the existing `inFence`/`inInline` tracking: it opens when an unescaped `](` is consumed outside code and closes at the first unescaped `)`. Markers inside that span are never recorded. - A link's `]` only counts toward bracket pairing once its URL closes, so hard truncation (4096-char limit) that slices mid-URL now leaves the `[` unmatched and trims back to before the link — previously the cut left an unterminated `(` that Telegram rejected. The defensive under-limit safety pass from #446 (streaming chunks) is intentionally kept; it's now link-aware. |
