szuperaz · Nov 10, 2026
@@ -388,7 +388,7 @@ const server = new StreamClient(
 const token = server.generateUserToken({ user_id: 'jane', validity_in_seconds: 60 * 60 });
 ```
 
-Client-side relatives: `client.revokeTokens(before?)` now revokes only the connected user's tokens (Ref: revoke-tokens); `client.unbanUser` keeps its signature but goes through V2 `/moderation/unban` (v1 only for `remove_future_channels_ban: true`).
+Client-side relatives: `client.revokeTokens(before?)` now revokes only the connected user's tokens (Ref: revoke-tokens); `client.unbanUser` keeps its signature but goes through V2 `/moderation/unban`.
 
 <a id="s2-client-construction-and-configuration"></a>
 

@@ -499,7 +499,7 @@ Common rules for every generated or rewritten `StreamChat` method: path paramete
 **Kind:** behavior · **Impact:** medium · **PR:** #1760
 
 - **v9:** endpoints at `baseURL + '/channels'`, `'/users'`, `'/messages/{id}'`, `'/uploads/file'`, ...; every request carried `user_id`, `api_key` and `connection_id`.
-- **v10:** generated endpoints are `baseURL + '/api/v2/...'` (e.g. `/api/v2/chat/channels`, `/api/v2/users`, `/api/v2/chat/messages/{id}`, `/api/v2/uploads/file`). `user_id` and `api_key` are still on every request; `connection_id` only on requests that watch or subscribe to presence and on `stopWatchingChannel`. Hand-written `unbanUser` calls V2 `/api/v2/moderation/unban`, except with `remove_future_channels_ban: true`, which still goes to v1 `DELETE /moderation/ban`.
+- **v10:** generated endpoints are `baseURL + '/api/v2/...'` (e.g. `/api/v2/chat/channels`, `/api/v2/users`, `/api/v2/chat/messages/{id}`, `/api/v2/uploads/file`). `user_id` and `api_key` are still on every request; `connection_id` only on requests that watch or subscribe to presence and on `stopWatchingChannel`. Hand-written `unbanUser` calls V2 `/api/v2/moderation/unban`.
 - **Migrate:** update reverse proxies, CSP allowlists, request mocks (msw / nock) and server-side request filters keyed on paths or on `connection_id`.
 - **Source:** `src/gen/chat/ChatApi.ts`, `src/api-client.ts:146-171,206-217`, `src/client.ts:1840` (v10); `src/client.ts:3929-3936` (v9)
 
@@ -1582,7 +1582,7 @@ Common pattern for every generated or overridden `Channel` method: the trailing
 **Kind:** signature · **Impact:** medium
 
 - **v9:** `banUser(targetUserID, options: BanUserOptions)` (v1 `/moderation/ban` with `type` / `id`), `unbanUser(targetUserID, options?)`, `shadowBan(targetUserID, options)`, `removeShadowBan(targetUserID)`; all required initialization.
-- **v10:** `banUser(targetUserId, options: Omit<BanUserOptions, 'channel_cid'>)` goes through `client.moderation.ban({ ...options, target_user_id, channel_cid })` (V2; returns `StreamResponse<ModerationBanResponse>`; `delete_reactions`, `ban_from_future_channels` follow `BanRequest`); `unbanUser(targetUserId, options?)` keeps its signature and goes through V2 `client.moderation.unban({ target_user_id, channel_cid })` (v1 only for `remove_future_channels_ban: true`). Both require only an id. `shadowBan` / `removeShadowBan` removed.
+- **v10:** `banUser(targetUserId, options: Omit<BanUserOptions, 'channel_cid'>)` goes through `client.moderation.ban({ ...options, target_user_id, channel_cid })` (V2; returns `StreamResponse<ModerationBanResponse>`; `delete_reactions`, `ban_from_future_channels` follow `BanRequest`); `unbanUser(targetUserId, options?)` keeps its signature and goes through V2 `client.moderation.unban({ target_user_id, channel_cid, remove_future_channels_ban })`. With a channel, `remove_future_channels_ban` removes the channel creator's future channels ban (where the V2 ban keys it) rather than the caller's. Both require only an id. `shadowBan` / `removeShadowBan` removed.
 - **Migrate:** `channel.shadowBan(id, o)` -> `channel.banUser(id, { ...o, shadow: true })`; `channel.removeShadowBan(id)` -> `channel.unbanUser(id, { shadow: true })`.
 - **Source:** `src/channel.ts:2094-2144`, `src/types.ts:97,139-147` (v10); `src/channel.ts:1711-1800` (v9)
 

@@ -826,7 +826,7 @@ Non-breaking; land any time. Breaking simplifications are in [section 1](#fix-be
 - **Where:** `src/client.ts:1839-1844`, `src/types.ts:59-68` (`APIResponse`)
 - **What:** `unbanUser` still calls v1 `DELETE /moderation/ban` returning the legacy `APIResponse` envelope, while ban, mute and unmute moved to V2 `client.moderation.*`, which already has `unban`. `APIResponse`'s own comment says to delete it with these methods. RN V10 calls `client.unbanUser(id)` (`useMessageActionHandlers.ts:216`).
 - **Suggested change:** Route through `this.moderation.unban({ target_user_id, ...options })` while keeping the `unbanUser(targetUserID, options?)` signature (non-breaking, any time). Deleting `APIResponse` is a v9 break, decided with the [restated types](#lo-restated-types).
-- **Status:** mostly done. `unbanUser` now calls `this.moderation.unban({ target_user_id, channel_cid })`, with `channel_cid` built from `type` / `id`. The exception is `remove_future_channels_ban: true`, which still goes to v1: the V2 `/unban` controller forwards only `channel_cid` to its `UnbanAction`, although that action supports the flag. The v1 path can go once the backend accepts the flag on V2. `APIResponse` stays, as the return type.
+- **Status:** done. `unbanUser` calls `this.moderation.unban({ target_user_id, channel_cid, remove_future_channels_ban })`, with `channel_cid` built from `type` / `id`; the v1 fallback for `remove_future_channels_ban` went once the spec gained the flag on V2. With a channel, V2 removes the channel creator's future channels ban (where the V2 ban keys it), where v1 removed the caller's, so a non-creator moderator's unban now actually removes it. `APIResponse` stays, as the return type.
 
 <a id="offline-cache-key"></a>
 

@@ -1683,10 +1683,11 @@ export class StreamChat extends ChatApi {
   /**
    * Revokes a global ban, or a channel ban when `options.type` and `options.id` name a channel.
    *
-   * Goes through V2 `client.moderation.unban()`. The one exception is
-   * `remove_future_channels_ban`, which that endpoint does not accept, so a call with it set still
-   * goes to v1 `DELETE /moderation/ban`. `shadow` is accepted for compatibility but ignored: the
-   * server lifts whichever ban is in place, shadow or not, on both endpoints.
+   * Goes through V2 `client.moderation.unban()`. `shadow` is accepted for compatibility but ignored:
+   * the server lifts whichever ban is in place, shadow or not. With `remove_future_channels_ban`,
+   * the future channels ban is removed too: the channel creator's when a channel is named (where
+   * `channel.banUser` with `ban_from_future_channels` keys it), otherwise the caller's, and then
+   * only the future channels ban is removed.
    *
    * @param targetUserId - The user to unban.
    * @param options - Unban options (optional).
@@ -1697,18 +1698,11 @@ export class StreamChat extends ChatApi {
     options?: UnBanUserOptions,
   ): Promise<APIResponse> {
     const { id, remove_future_channels_ban, type } = options ?? {};
-    const target_user_id = options?.target_user_id ?? targetUserId;
-
-    if (remove_future_channels_ban) {
-      return await this.api.delete<APIResponse>(this.baseURL + '/moderation/ban', {
-        ...options,
-        target_user_id,
-      });
-    }
 
     return await this.moderation.unban({
-      target_user_id,
+      target_user_id: options?.target_user_id ?? targetUserId,
       channel_cid: type && id ? `${type}:${id}` : undefined,
+      remove_future_channels_ban,
     });
   }
 

@@ -101,6 +101,8 @@ import type {
   Response,
   RewindChannelRequest,
   RewindChannelResponse,
+  SearchAllRequest,
+  SearchAllResponse,
   SearchPayload,
   SearchResponse,
   SearchRolesResponse,
@@ -1269,6 +1271,24 @@ export class ChatApi {
     );
   }
 
+  searchAll(
+    request?: SearchAllRequest & { connection_id?: string },
+    requestOptions?: StreamRequestOptions,
+  ): Promise<StreamResponse<SearchAllResponse>> {
+    const { connection_id, ...body } = request ?? {};
+    const queryParams = { connection_id };
+
+    return this.apiClient.sendRequest<SearchAllResponse>(
+      'POST',
+      '/api/v2/chat/search/all',
+      undefined,
+      queryParams,
+      body,
+      'application/json',
+      requestOptions,
+    );
+  }
+
   sync(
     request: SyncRequest & {
       with_inaccessible_cids?: boolean;

@@ -2573,6 +2573,7 @@ export interface ClosedCaptionRuleParameters {
   threshold?: number;
   time_window?: string;
   harm_labels?: Array<string>;
+  severities?: Array<string>;
   llm_harm_labels?: Record<string, string>;
 }
 
@@ -5878,11 +5879,15 @@ export interface ModerationEngineConfigRequest {
   /**
    * LLM model used when the provider resolves to llm. Not allowed with provider nlp. Empty uses the app's moderation settings
    */
-  model?: 'gpt-4o-mini' | 'gpt-4o';
+  model?: 'gpt-4o-mini' | 'gpt-4o' | 'gpt-6-luna';
   /**
    * llm or nlp. Empty follows the app's LLM moderation setting
    */
   provider?: 'llm' | 'nlp';
+  /**
+   * OpenAI reasoning effort for model. Requires model, and must be one the model accepts. Empty uses the model's default
+   */
+  reasoning_effort?: string;
 }
 
 export interface ModerationEngineConfigResponse {
@@ -5894,6 +5899,10 @@ export interface ModerationEngineConfigResponse {
    * llm or nlp. Empty follows the app's LLM moderation setting
    */
   provider?: 'llm' | 'nlp';
+  /**
+   * OpenAI reasoning effort for model. Requires model, and must be one the model accepts. Empty uses the model's default
+   */
+  reasoning_effort?: string;
 }
 
 export interface ModerationFlagResponse {
@@ -10073,6 +10082,93 @@ export interface ScheduledMessagesFilter {
   status?: Array<ScheduledMessageQueryStatus>;
 }
 
+export interface SearchAllChannelsSection {
+  /**
+   * Opaque cursor for the next page of this section, present only when more results exist. Pass it back under next.channels with the same query
+   */
+  next?: string;
+  /**
+   * Channels the caller is a member of whose name starts with the query, in the same shape as the query channels endpoint's results but without messages or members
+   */
+  channels?: Array<ChannelStateResponseFields>;
+  error?: SearchAllSectionError;
+  warning?: SearchAllWarning;
+}
+
+export interface SearchAllMessagesSection {
+  /**
+   * Opaque cursor for the next page of this section, present only when more results exist. Pass it back under next.messages with the same query
+   */
+  next?: string;
+  /**
+   * Matching messages, newest first, in the same shape as the search endpoint's results
+   */
+  results?: Array<SearchResult>;
+  error?: SearchAllSectionError;
+  results_warning?: SearchWarning;
+  warning?: SearchAllWarning;
+}
+
+export interface SearchAllRequest {
+  /**
+   * Number of results per section, 1-30. Defaults to 10. Applies to every requested section; there is no per-section limit
+   */
+  limit?: number;
+  /**
+   * Free-text search phrase, 1-80 characters after trimming whitespace. Required. Matched as a prefix of a user's name or username and of a channel's name, and as full-text against message text
+   */
+  query?: string;
+  /**
+   * Which sections to search: any subset of messages, users and channels. Defaults to all three when omitted; must not be empty when provided. Response sections always appear in the order messages, users, channels
+   */
+  entities?: Array<string>;
+  /**
+   * Opaque pagination cursors from a previous response, keyed by entity. A cursor that fails to decode restarts that section from page 1 with a cursor_rejected warning; the request fails with 400 only when every cursor supplied for a requested entity is invalid. Cursors for entities not in entities are ignored
+   */
+  next?: Record<string, string>;
+}
+
+export interface SearchAllResponse {
+  /**
+   * Duration of the request in milliseconds
+   */
+  duration: string;
+  channels?: SearchAllChannelsSection;
+  messages?: SearchAllMessagesSection;
+  users?: SearchAllUsersSection;
+}
+
+export interface SearchAllSectionError {
+  /**
+   * Error code. backend_timeout and backend_unavailable are retryable; unsupported (the app's configuration cannot serve this section) and internal_error are not
+   */
+  code: 'backend_unavailable' | 'backend_timeout' | 'unsupported' | 'internal_error';
+  /**
+   * Human-readable description of the failure
+   */
+  message: string;
+}
+
+export interface SearchAllUsersSection {
+  /**
+   * Opaque cursor for the next page of this section, present only when more results exist. Pass it back under next.users with the same query
+   */
+  next?: string;
+  /**
+   * Users whose name or username starts with the query, in the same shape as the query users endpoint's results
+   */
+  users?: Array<FullUserResponse>;
+  error?: SearchAllSectionError;
+  warning?: SearchAllWarning;
+}
+
+export interface SearchAllWarning {
+  /**
+   * Warning code. cursor_rejected: the cursor supplied for this section failed to decode, so the section restarted from page 1
+   */
+  code: 'cursor_rejected';
+}
+
 export interface SearchPayload {
   /**
    * Channel filter conditions

@@ -489,11 +489,15 @@ export class ModerationApi {
   }
 
   unban(
-    request: UnbanRequest & { target_user_id: string; channel_cid?: string },
+    request: UnbanRequest & {
+      target_user_id: string;
+      channel_cid?: string;
+      remove_future_channels_ban?: boolean;
+    },
     requestOptions?: StreamRequestOptions,
   ): Promise<StreamResponse<UnbanResponse>> {
-    const { target_user_id, channel_cid, ...body } = request;
-    const queryParams = { target_user_id, channel_cid };
+    const { target_user_id, channel_cid, remove_future_channels_ban, ...body } = request;
+    const queryParams = { target_user_id, channel_cid, remove_future_channels_ban };
 
     return this.apiClient.sendRequest<UnbanResponse>(
       'POST',

@@ -51,8 +51,8 @@ export type RequireAtLeastOne<T> = {
 }[keyof T];
 
 /**
- * The return type of `StreamChat.unbanUser`, kept so its signature is unchanged. Its v1 fallback
- * (`remove_future_channels_ban`) is the one hand-written `/moderation/*` call left.
+ * The return type of `StreamChat.unbanUser`, kept so its signature is unchanged. `unbanUser` itself
+ * goes through the generated V2 `moderation.unban()`.
  */
 export type APIResponse = {
   duration: string;

@@ -56,13 +56,13 @@ describe('unbanUser', () => {
         shadow: true,
         client_id: 'c',
         connection_id: 'x',
-        remove_future_channels_ban: false,
       });
 
       expect(unbanCall()).toMatchObject({
         queryParams: { target_user_id: 'bob', channel_cid: undefined },
         body: {},
       });
+      expect(unbanCall().queryParams).not.toHaveProperty('shadow');
     });
 
     it('keeps options.target_user_id taking precedence, as before', async () => {
@@ -71,14 +71,19 @@ describe('unbanUser', () => {
       expect(unbanCall().queryParams).toMatchObject({ target_user_id: 'alice' });
     });
 
-    it('falls back to v1 for remove_future_channels_ban, which V2 does not accept', async () => {
-      await client.unbanUser('bob', { remove_future_channels_ban: true });
-
-      expect(sendRequest).not.toHaveBeenCalled();
-      expect(legacyDelete).toHaveBeenCalledWith(`${client.baseURL}/moderation/ban`, {
+    it('sends remove_future_channels_ban to the V2 endpoint', async () => {
+      await client.unbanUser('bob', {
+        type: 'messaging',
+        id: 'general',
         remove_future_channels_ban: true,
+      });
+
+      expect(unbanCall().queryParams).toEqual({
         target_user_id: 'bob',
+        channel_cid: 'messaging:general',
+        remove_future_channels_ban: true,
       });
+      expect(legacyDelete).not.toHaveBeenCalled();
     });
   });