fix(slack): enforce replyToMode for auto-thread_ts and inline reply tags (#23839)

* Slack: respect replyToMode for auto-thread_ts and inline reply tags

* Update CHANGELOG.md

src/slack/monitor/message-handler/dispatch.ts

@@ -103,7 +103,6 @@ export async function dispatchPreparedSlackMessage(prepared: PreparedSlackMessag
     incomingThreadTs,
     messageTs,
     hasRepliedRef,
-    chatType: prepared.isDirectMessage ? "direct" : "channel",
     isThreadReply,
   });
 
@@ -187,6 +186,7 @@ export async function dispatchPreparedSlackMessage(prepared: PreparedSlackMessag
       runtime,
       textLimit: ctx.textLimit,
       replyThreadTs,
+      replyToMode: ctx.replyToMode,
     });
     replyPlan.markSent();
   };

src/slack/monitor/replies.ts

@@ -16,9 +16,13 @@ export async function deliverReplies(params: {
   runtime: RuntimeEnv;
   textLimit: number;
   replyThreadTs?: string;
+  replyToMode: "off" | "first" | "all";
 }) {
   for (const payload of params.replies) {
-    const threadTs = payload.replyToId ?? params.replyThreadTs;
+    // Keep reply tags opt-in: when replyToMode is off, explicit reply tags
+    // must not force threading.
+    const inlineReplyToId = params.replyToMode === "off" ? undefined : payload.replyToId;
+    const threadTs = inlineReplyToId ?? params.replyThreadTs;
     const mediaList = payload.mediaUrls ?? (payload.mediaUrl ? [payload.mediaUrl] : []);
     const text = payload.text ?? "";
     if (!text && mediaList.length === 0) {
@@ -88,19 +92,11 @@ function createSlackReplyReferencePlanner(params: {
   incomingThreadTs: string | undefined;
   messageTs: string | undefined;
   hasReplied?: boolean;
-  chatType?: "direct" | "channel" | "group";
   isThreadReply?: boolean;
 }) {
-  // When already inside a Slack thread, stay in it — but for DMs where the
-  // "thread" was created by the typing indicator (not a real thread reply),
-  // respect the user's replyToMode setting.
-  // See: https://github.com/openclaw/openclaw/issues/16868
-  const effectiveMode =
-    params.chatType === "direct" && !params.isThreadReply
-      ? params.replyToMode
-      : params.incomingThreadTs
-        ? "all"
-        : params.replyToMode;
+  // Only force threading for real user thread replies. If Slack auto-populates
+  // thread_ts on top-level messages, preserve the configured reply mode.
+  const effectiveMode = params.isThreadReply ? "all" : params.replyToMode;
   return createReplyReferencePlanner({
     replyToMode: effectiveMode,
     existingId: params.incomingThreadTs,
@@ -114,7 +110,6 @@ export function createSlackReplyDeliveryPlan(params: {
   incomingThreadTs: string | undefined;
   messageTs: string | undefined;
   hasRepliedRef: { value: boolean };
-  chatType?: "direct" | "channel" | "group";
   isThreadReply?: boolean;
 }): SlackReplyDeliveryPlan {
   const replyReference = createSlackReplyReferencePlanner({
@@ -122,7 +117,6 @@ export function createSlackReplyDeliveryPlan(params: {
     incomingThreadTs: params.incomingThreadTs,
     messageTs: params.messageTs,
     hasReplied: params.hasRepliedRef.value,
-    chatType: params.chatType,
     isThreadReply: params.isThreadReply,
   });
   return {