fix: preserve block streaming order

CHANGELOG.md

@@ -15,6 +15,7 @@
 ### Fixes
 - Telegram: chunk block-stream replies to avoid “message is too long” errors (#124) — thanks @mukhtharcm.
 - Block streaming: default to text_end and suppress duplicate block sends while in-flight.
+- Block streaming: drop final payloads after soft chunking to keep Discord order intact.
 - Gmail hooks: resolve gcloud Python to a real executable when PATH uses mise shims — thanks @joargp.
 - Control UI: generate UUIDs when `crypto.randomUUID()` is unavailable over HTTP — thanks @ratulsarna.
 - Agent: add soft block-stream chunking (800–1200 chars default) with paragraph/newline preference.

src/auto-reply/reply.ts

@@ -1133,6 +1133,7 @@ export async function getReplyFromConfig(
   const streamedPayloadKeys = new Set<string>();
   const pendingStreamedPayloadKeys = new Set<string>();
   const pendingBlockTasks = new Set<Promise<void>>();
+  let didStreamBlockReply = false;
   const buildPayloadKey = (payload: ReplyPayload) => {
     const text = payload.text?.trim() ?? "";
     const mediaList = payload.mediaUrls?.length
@@ -2239,6 +2240,7 @@ export async function getReplyFromConfig(
                   return;
                 }
                 pendingStreamedPayloadKeys.add(payloadKey);
+                didStreamBlockReply = true;
                 const task = (async () => {
                   await startTypingOnText(cleaned);
                   await opts.onBlockReply?.(blockPayload);
@@ -2345,11 +2347,15 @@ export async function getReplyFromConfig(
           (payload.mediaUrls && payload.mediaUrls.length > 0),
       );
 
-    const filteredPayloads = blockStreamingEnabled
-      ? replyTaggedPayloads.filter(
-          (payload) => !streamedPayloadKeys.has(buildPayloadKey(payload)),
-        )
-      : replyTaggedPayloads;
+    const shouldDropFinalPayloads =
+      blockStreamingEnabled && blockReplyChunking && didStreamBlockReply;
+    const filteredPayloads = shouldDropFinalPayloads
+      ? []
+      : blockStreamingEnabled
+        ? replyTaggedPayloads.filter(
+            (payload) => !streamedPayloadKeys.has(buildPayloadKey(payload)),
+          )
+        : replyTaggedPayloads;
 
     if (filteredPayloads.length === 0) return finalizeWithFollowup(undefined);