diff --git a/CHANGELOG.md b/CHANGELOG.md
index e88e5f1f2..55ae725a5 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -55,6 +55,7 @@
 ### Fixes
 - Chat UI: keep the chat scrolled to the latest message after switching sessions.
 - Auto-reply: stream completed reply blocks as soon as they finish (configurable default + break); skip empty tool-only blocks unless verbose.
+- Messages: make outbound text chunk limits configurable (defaults remain 4000/Discord 2000).
 - CLI onboarding: persist gateway token in config so local CLI auth works; recommend auth Off unless you need multi-machine access.
 - Control UI: accept a `?token=` URL param to auto-fill Gateway auth; onboarding now opens the dashboard with token auth when configured.
 - Agent prompt: remove hardcoded user name in system prompt example.
diff --git a/docs/configuration.md b/docs/configuration.md
index 2c1506838..4d7db7238 100644
--- a/docs/configuration.md
+++ b/docs/configuration.md
@@ -282,7 +282,17 @@ Controls inbound/outbound prefixes and timestamps.
   messages: {
     messagePrefix: "[clawdis]",
     responsePrefix: "🦞",
-    timestampPrefix: "Europe/London"
+    timestampPrefix: "Europe/London",
+    // outbound chunk size (chars); defaults vary by surface (e.g. 4000, Discord 2000)
+    textChunkLimit: 4000,
+    // optional per-surface overrides
+    textChunkLimitBySurface: {
+      whatsapp: 4000,
+      telegram: 4000,
+      signal: 4000,
+      imessage: 4000,
+      discord: 2000
+    }
   }
 }
 ```
diff --git a/src/auto-reply/chunk.test.ts b/src/auto-reply/chunk.test.ts
index 065edc672..d0b336112 100644
--- a/src/auto-reply/chunk.test.ts
+++ b/src/auto-reply/chunk.test.ts
@@ -1,6 +1,6 @@
 import { describe, expect, it } from "vitest";
 
-import { chunkText } from "./chunk.js";
+import { chunkText, resolveTextChunkLimit } from "./chunk.js";
 
 describe("chunkText", () => {
   it("keeps multi-line text in one chunk when under limit", () => {
@@ -45,3 +45,30 @@ describe("chunkText", () => {
     expect(chunks).toEqual(["Supercalif", "ragilistic", "expialidoc", "ious"]);
   });
 });
+
+describe("resolveTextChunkLimit", () => {
+  it("uses per-surface defaults", () => {
+    expect(resolveTextChunkLimit(undefined, "whatsapp")).toBe(4000);
+    expect(resolveTextChunkLimit(undefined, "telegram")).toBe(4000);
+    expect(resolveTextChunkLimit(undefined, "signal")).toBe(4000);
+    expect(resolveTextChunkLimit(undefined, "imessage")).toBe(4000);
+    expect(resolveTextChunkLimit(undefined, "discord")).toBe(2000);
+  });
+
+  it("supports a global override", () => {
+    const cfg = { messages: { textChunkLimit: 1234 } };
+    expect(resolveTextChunkLimit(cfg, "whatsapp")).toBe(1234);
+    expect(resolveTextChunkLimit(cfg, "discord")).toBe(1234);
+  });
+
+  it("prefers per-surface overrides over global", () => {
+    const cfg = {
+      messages: {
+        textChunkLimit: 1234,
+        textChunkLimitBySurface: { discord: 111 },
+      },
+    };
+    expect(resolveTextChunkLimit(cfg, "discord")).toBe(111);
+    expect(resolveTextChunkLimit(cfg, "telegram")).toBe(1234);
+  });
+});
diff --git a/src/auto-reply/chunk.ts b/src/auto-reply/chunk.ts
index a9470db87..caac3feae 100644
--- a/src/auto-reply/chunk.ts
+++ b/src/auto-reply/chunk.ts
@@ -2,6 +2,43 @@
 // unintentionally breaking on newlines. Using [\s\S] keeps newlines inside
 // the chunk so messages are only split when they truly exceed the limit.
 
+import type { ClawdisConfig } from "../config/config.js";
+
+export type TextChunkSurface =
+  | "whatsapp"
+  | "telegram"
+  | "discord"
+  | "signal"
+  | "imessage"
+  | "webchat";
+
+const DEFAULT_CHUNK_LIMIT_BY_SURFACE: Record<TextChunkSurface, number> = {
+  whatsapp: 4000,
+  telegram: 4000,
+  discord: 2000,
+  signal: 4000,
+  imessage: 4000,
+  webchat: 4000,
+};
+
+export function resolveTextChunkLimit(
+  cfg: Pick<ClawdisConfig, "messages"> | undefined,
+  surface?: TextChunkSurface,
+): number {
+  const surfaceOverride = surface
+    ? cfg?.messages?.textChunkLimitBySurface?.[surface]
+    : undefined;
+  if (typeof surfaceOverride === "number" && surfaceOverride > 0) {
+    return surfaceOverride;
+  }
+  const globalOverride = cfg?.messages?.textChunkLimit;
+  if (typeof globalOverride === "number" && globalOverride > 0) {
+    return globalOverride;
+  }
+  if (surface) return DEFAULT_CHUNK_LIMIT_BY_SURFACE[surface];
+  return 4000;
+}
+
 export function chunkText(text: string, limit: number): string[] {
   if (!text) return [];
   if (limit <= 0) return [text];
diff --git a/src/commands/agent.ts b/src/commands/agent.ts
index d62398f27..1518f70a1 100644
--- a/src/commands/agent.ts
+++ b/src/commands/agent.ts
@@ -17,7 +17,7 @@ import {
   DEFAULT_AGENT_WORKSPACE_DIR,
   ensureAgentWorkspace,
 } from "../agents/workspace.js";
-import { chunkText } from "../auto-reply/chunk.js";
+import { chunkText, resolveTextChunkLimit } from "../auto-reply/chunk.js";
 import type { MsgContext } from "../auto-reply/templating.js";
 import {
   normalizeThinkLevel,
@@ -524,6 +524,15 @@ export async function agentCommand(
     return;
   }
 
+  const deliveryTextLimit =
+    deliveryProvider === "whatsapp" ||
+    deliveryProvider === "telegram" ||
+    deliveryProvider === "discord" ||
+    deliveryProvider === "signal" ||
+    deliveryProvider === "imessage"
+      ? resolveTextChunkLimit(cfg, deliveryProvider)
+      : resolveTextChunkLimit(cfg, "whatsapp");
+
   for (const payload of payloads) {
     const mediaList =
       payload.mediaUrls ?? (payload.mediaUrl ? [payload.mediaUrl] : []);
@@ -564,7 +573,7 @@ export async function agentCommand(
     if (deliveryProvider === "telegram" && telegramTarget) {
       try {
         if (media.length === 0) {
-          for (const chunk of chunkText(text, 4000)) {
+          for (const chunk of chunkText(text, deliveryTextLimit)) {
             await deps.sendMessageTelegram(telegramTarget, chunk, {
               verbose: false,
               token: telegramToken || undefined,
@@ -645,7 +654,7 @@ export async function agentCommand(
     if (deliveryProvider === "imessage" && imessageTarget) {
       try {
         if (media.length === 0) {
-          for (const chunk of chunkText(text, 4000)) {
+          for (const chunk of chunkText(text, deliveryTextLimit)) {
             await deps.sendMessageIMessage(imessageTarget, chunk, {
               maxBytes: cfg.imessage?.mediaMaxMb
                 ? cfg.imessage.mediaMaxMb * 1024 * 1024
diff --git a/src/config/config.ts b/src/config/config.ts
index 1a1dc31b1..1a9b1eca2 100644
--- a/src/config/config.ts
+++ b/src/config/config.ts
@@ -314,6 +314,15 @@ export type MessagesConfig = {
   messagePrefix?: string; // Prefix added to all inbound messages (default: "[clawdis]" if no allowFrom, else "")
   responsePrefix?: string; // Prefix auto-added to all outbound replies (e.g., "🦞")
   timestampPrefix?: boolean | string; // true/false or IANA timezone string (default: true with UTC)
+  /** Outbound text chunk size (chars). Default varies by provider (e.g. 4000, Discord 2000). */
+  textChunkLimit?: number;
+  /** Optional per-surface chunk overrides. */
+  textChunkLimitBySurface?: Partial<
+    Record<
+      "whatsapp" | "telegram" | "discord" | "signal" | "imessage" | "webchat",
+      number
+    >
+  >;
 };
 
 export type BridgeBindMode = "auto" | "lan" | "tailnet" | "loopback";
@@ -708,6 +717,17 @@ const MessagesSchema = z
     messagePrefix: z.string().optional(),
     responsePrefix: z.string().optional(),
     timestampPrefix: z.union([z.boolean(), z.string()]).optional(),
+    textChunkLimit: z.number().int().positive().optional(),
+    textChunkLimitBySurface: z
+      .object({
+        whatsapp: z.number().int().positive().optional(),
+        telegram: z.number().int().positive().optional(),
+        discord: z.number().int().positive().optional(),
+        signal: z.number().int().positive().optional(),
+        imessage: z.number().int().positive().optional(),
+        webchat: z.number().int().positive().optional(),
+      })
+      .optional(),
   })
   .optional();
 
diff --git a/src/cron/isolated-agent.ts b/src/cron/isolated-agent.ts
index 2ed808e91..e972837d3 100644
--- a/src/cron/isolated-agent.ts
+++ b/src/cron/isolated-agent.ts
@@ -12,7 +12,7 @@ import {
   DEFAULT_AGENT_WORKSPACE_DIR,
   ensureAgentWorkspace,
 } from "../agents/workspace.js";
-import { chunkText } from "../auto-reply/chunk.js";
+import { chunkText, resolveTextChunkLimit } from "../auto-reply/chunk.js";
 import { normalizeThinkLevel } from "../auto-reply/thinking.js";
 import type { CliDeps } from "../cli/deps.js";
 import type { ClawdisConfig } from "../config/config.js";
@@ -357,12 +357,13 @@ export async function runCronIsolatedAgentTurn(params: {
         };
       }
       const chatId = resolvedDelivery.to;
+      const textLimit = resolveTextChunkLimit(params.cfg, "telegram");
       try {
         for (const payload of payloads) {
           const mediaList =
             payload.mediaUrls ?? (payload.mediaUrl ? [payload.mediaUrl] : []);
           if (mediaList.length === 0) {
-            for (const chunk of chunkText(payload.text ?? "", 4000)) {
+            for (const chunk of chunkText(payload.text ?? "", textLimit)) {
               await params.deps.sendMessageTelegram(chatId, chunk, {
                 verbose: false,
                 token: telegramToken || undefined,
@@ -444,12 +445,13 @@ export async function runCronIsolatedAgentTurn(params: {
         };
       }
       const to = resolvedDelivery.to;
+      const textLimit = resolveTextChunkLimit(params.cfg, "signal");
       try {
         for (const payload of payloads) {
           const mediaList =
             payload.mediaUrls ?? (payload.mediaUrl ? [payload.mediaUrl] : []);
           if (mediaList.length === 0) {
-            for (const chunk of chunkText(payload.text ?? "", 4000)) {
+            for (const chunk of chunkText(payload.text ?? "", textLimit)) {
               await params.deps.sendMessageSignal(to, chunk);
             }
           } else {
@@ -482,12 +484,13 @@ export async function runCronIsolatedAgentTurn(params: {
         };
       }
       const to = resolvedDelivery.to;
+      const textLimit = resolveTextChunkLimit(params.cfg, "imessage");
       try {
         for (const payload of payloads) {
           const mediaList =
             payload.mediaUrls ?? (payload.mediaUrl ? [payload.mediaUrl] : []);
           if (mediaList.length === 0) {
-            for (const chunk of chunkText(payload.text ?? "", 4000)) {
+            for (const chunk of chunkText(payload.text ?? "", textLimit)) {
               await params.deps.sendMessageIMessage(to, chunk);
             }
           } else {
diff --git a/src/imessage/monitor.ts b/src/imessage/monitor.ts
index 96e4c8315..281c62830 100644
--- a/src/imessage/monitor.ts
+++ b/src/imessage/monitor.ts
@@ -1,4 +1,4 @@
-import { chunkText } from "../auto-reply/chunk.js";
+import { chunkText, resolveTextChunkLimit } from "../auto-reply/chunk.js";
 import { formatAgentEnvelope } from "../auto-reply/envelope.js";
 import { getReplyFromConfig } from "../auto-reply/reply.js";
 import type { ReplyPayload } from "../auto-reply/types.js";
@@ -111,15 +111,16 @@ async function deliverReplies(params: {
   client: Awaited<ReturnType<typeof createIMessageRpcClient>>;
   runtime: RuntimeEnv;
   maxBytes: number;
+  textLimit: number;
 }) {
-  const { replies, target, client, runtime, maxBytes } = params;
+  const { replies, target, client, runtime, maxBytes, textLimit } = params;
   for (const payload of replies) {
     const mediaList =
       payload.mediaUrls ?? (payload.mediaUrl ? [payload.mediaUrl] : []);
     const text = payload.text ?? "";
     if (!text && mediaList.length === 0) continue;
     if (mediaList.length === 0) {
-      for (const chunk of chunkText(text, 4000)) {
+      for (const chunk of chunkText(text, textLimit)) {
         await sendMessageIMessage(target, chunk, { maxBytes, client });
       }
     } else {
@@ -143,6 +144,7 @@ export async function monitorIMessageProvider(
 ): Promise<void> {
   const runtime = resolveRuntime(opts);
   const cfg = loadConfig();
+  const textLimit = resolveTextChunkLimit(cfg, "imessage");
   const allowFrom = resolveAllowFrom(opts);
   const mentionRegexes = resolveMentionRegexes(cfg);
   const includeAttachments =
@@ -274,6 +276,7 @@ export async function monitorIMessageProvider(
             client,
             runtime,
             maxBytes: mediaMaxBytes,
+            textLimit,
           });
         })
         .catch((err) => {
@@ -302,6 +305,7 @@ export async function monitorIMessageProvider(
       client,
       runtime,
       maxBytes: mediaMaxBytes,
+      textLimit,
     });
   };
 
diff --git a/src/infra/heartbeat-runner.ts b/src/infra/heartbeat-runner.ts
index 822f537ee..d3e1ce383 100644
--- a/src/infra/heartbeat-runner.ts
+++ b/src/infra/heartbeat-runner.ts
@@ -1,4 +1,4 @@
-import { chunkText } from "../auto-reply/chunk.js";
+import { chunkText, resolveTextChunkLimit } from "../auto-reply/chunk.js";
 import {
   HEARTBEAT_PROMPT,
   stripHeartbeatToken,
@@ -292,6 +292,7 @@ async function deliverHeartbeatReply(params: {
   to: string;
   text: string;
   mediaUrls: string[];
+  textLimit: number;
   deps: Required<
     Pick<
       HeartbeatDeps,
@@ -303,10 +304,10 @@ async function deliverHeartbeatReply(params: {
     >
   >;
 }) {
-  const { channel, to, text, mediaUrls, deps } = params;
+  const { channel, to, text, mediaUrls, deps, textLimit } = params;
   if (channel === "whatsapp") {
     if (mediaUrls.length === 0) {
-      for (const chunk of chunkText(text, 4000)) {
+      for (const chunk of chunkText(text, textLimit)) {
         await deps.sendWhatsApp(to, chunk, { verbose: false });
       }
       return;
@@ -322,7 +323,7 @@ async function deliverHeartbeatReply(params: {
 
   if (channel === "signal") {
     if (mediaUrls.length === 0) {
-      for (const chunk of chunkText(text, 4000)) {
+      for (const chunk of chunkText(text, textLimit)) {
         await deps.sendSignal(to, chunk);
       }
       return;
@@ -338,7 +339,7 @@ async function deliverHeartbeatReply(params: {
 
   if (channel === "imessage") {
     if (mediaUrls.length === 0) {
-      for (const chunk of chunkText(text, 4000)) {
+      for (const chunk of chunkText(text, textLimit)) {
         await deps.sendIMessage(to, chunk);
       }
       return;
@@ -354,7 +355,7 @@ async function deliverHeartbeatReply(params: {
 
   if (channel === "telegram") {
     if (mediaUrls.length === 0) {
-      for (const chunk of chunkText(text, 4000)) {
+      for (const chunk of chunkText(text, textLimit)) {
         await deps.sendTelegram(to, chunk, { verbose: false });
       }
       return;
@@ -500,11 +501,13 @@ export async function runHeartbeatOnce(opts: {
       sendSignal: opts.deps?.sendSignal ?? sendMessageSignal,
       sendIMessage: opts.deps?.sendIMessage ?? sendMessageIMessage,
     };
+    const textLimit = resolveTextChunkLimit(cfg, delivery.channel);
     await deliverHeartbeatReply({
       channel: delivery.channel,
       to: delivery.to,
       text: normalized.text,
       mediaUrls,
+      textLimit,
       deps,
     });
 
diff --git a/src/signal/monitor.ts b/src/signal/monitor.ts
index 1f6d443ae..d7917d49b 100644
--- a/src/signal/monitor.ts
+++ b/src/signal/monitor.ts
@@ -1,4 +1,4 @@
-import { chunkText } from "../auto-reply/chunk.js";
+import { chunkText, resolveTextChunkLimit } from "../auto-reply/chunk.js";
 import { formatAgentEnvelope } from "../auto-reply/envelope.js";
 import { getReplyFromConfig } from "../auto-reply/reply.js";
 import type { ReplyPayload } from "../auto-reply/types.js";
@@ -178,15 +178,17 @@ async function deliverReplies(params: {
   account?: string;
   runtime: RuntimeEnv;
   maxBytes: number;
+  textLimit: number;
 }) {
-  const { replies, target, baseUrl, account, runtime, maxBytes } = params;
+  const { replies, target, baseUrl, account, runtime, maxBytes, textLimit } =
+    params;
   for (const payload of replies) {
     const mediaList =
       payload.mediaUrls ?? (payload.mediaUrl ? [payload.mediaUrl] : []);
     const text = payload.text ?? "";
     if (!text && mediaList.length === 0) continue;
     if (mediaList.length === 0) {
-      for (const chunk of chunkText(text, 4000)) {
+      for (const chunk of chunkText(text, textLimit)) {
         await sendMessageSignal(target, chunk, {
           baseUrl,
           account,
@@ -215,6 +217,7 @@ export async function monitorSignalProvider(
 ): Promise<void> {
   const runtime = resolveRuntime(opts);
   const cfg = loadConfig();
+  const textLimit = resolveTextChunkLimit(cfg, "signal");
   const baseUrl = resolveBaseUrl(opts);
   const account = resolveAccount(opts);
   const allowFrom = resolveAllowFrom(opts);
@@ -391,6 +394,7 @@ export async function monitorSignalProvider(
               account,
               runtime,
               maxBytes: mediaMaxBytes,
+              textLimit,
             });
           })
           .catch((err) => {
@@ -420,6 +424,7 @@ export async function monitorSignalProvider(
         account,
         runtime,
         maxBytes: mediaMaxBytes,
+        textLimit,
       });
     };
 
diff --git a/src/telegram/bot.ts b/src/telegram/bot.ts
index 7d88334f9..c5fbb6bc8 100644
--- a/src/telegram/bot.ts
+++ b/src/telegram/bot.ts
@@ -5,7 +5,7 @@ import { apiThrottler } from "@grammyjs/transformer-throttler";
 import type { ApiClientOptions, Message } from "grammy";
 import { Bot, InputFile, webhookCallback } from "grammy";
 
-import { chunkText } from "../auto-reply/chunk.js";
+import { chunkText, resolveTextChunkLimit } from "../auto-reply/chunk.js";
 import { formatAgentEnvelope } from "../auto-reply/envelope.js";
 import { getReplyFromConfig } from "../auto-reply/reply.js";
 import type { ReplyPayload } from "../auto-reply/types.js";
@@ -60,6 +60,7 @@ export function createTelegramBot(opts: TelegramBotOptions) {
   bot.api.config.use(apiThrottler());
 
   const cfg = loadConfig();
+  const textLimit = resolveTextChunkLimit(cfg, "telegram");
   const allowFrom = opts.allowFrom ?? cfg.telegram?.allowFrom;
   const replyToMode = opts.replyToMode ?? cfg.telegram?.replyToMode ?? "off";
   const mediaMaxBytes =
@@ -245,6 +246,7 @@ export function createTelegramBot(opts: TelegramBotOptions) {
         runtime,
         bot,
         replyToMode,
+        textLimit,
       });
     } catch (err) {
       runtime.error?.(danger(`handler failed: ${String(err)}`));
@@ -268,8 +270,9 @@ async function deliverReplies(params: {
   runtime: RuntimeEnv;
   bot: Bot;
   replyToMode: ReplyToMode;
+  textLimit: number;
 }) {
-  const { replies, chatId, runtime, bot, replyToMode } = params;
+  const { replies, chatId, runtime, bot, replyToMode, textLimit } = params;
   let hasReplied = false;
   for (const reply of replies) {
     if (!reply?.text && !reply?.mediaUrl && !(reply?.mediaUrls?.length ?? 0)) {
@@ -286,7 +289,7 @@ async function deliverReplies(params: {
         ? [reply.mediaUrl]
         : [];
     if (mediaList.length === 0) {
-      for (const chunk of chunkText(reply.text || "", 4000)) {
+      for (const chunk of chunkText(reply.text || "", textLimit)) {
         await sendTelegramText(bot, chatId, chunk, runtime, {
           replyToMessageId:
             replyToId && (replyToMode === "all" || !hasReplied)
diff --git a/src/web/auto-reply.ts b/src/web/auto-reply.ts
index 67be507a6..3d4879a99 100644
--- a/src/web/auto-reply.ts
+++ b/src/web/auto-reply.ts
@@ -1,4 +1,4 @@
-import { chunkText } from "../auto-reply/chunk.js";
+import { chunkText, resolveTextChunkLimit } from "../auto-reply/chunk.js";
 import { formatAgentEnvelope } from "../auto-reply/envelope.js";
 import {
   normalizeGroupActivation,
@@ -41,7 +41,6 @@ import {
 } from "./reconnect.js";
 import { formatError, getWebAuthAgeMs, readWebSelfId } from "./session.js";
 
-const WEB_TEXT_LIMIT = 4000;
 const DEFAULT_GROUP_HISTORY_LIMIT = 50;
 const whatsappLog = createSubsystemLogger("gateway/providers/whatsapp");
 const whatsappInboundLog = whatsappLog.child("inbound");
@@ -502,6 +501,7 @@ async function deliverWebReply(params: {
   replyResult: ReplyPayload;
   msg: WebInboundMsg;
   maxMediaBytes: number;
+  textLimit: number;
   replyLogger: ReturnType<typeof getChildLogger>;
   connectionId?: string;
   skipLog?: boolean;
@@ -510,12 +510,13 @@ async function deliverWebReply(params: {
     replyResult,
     msg,
     maxMediaBytes,
+    textLimit,
     replyLogger,
     connectionId,
     skipLog,
   } = params;
   const replyStarted = Date.now();
-  const textChunks = chunkText(replyResult.text || "", WEB_TEXT_LIMIT);
+  const textChunks = chunkText(replyResult.text || "", textLimit);
   const mediaList = replyResult.mediaUrls?.length
     ? replyResult.mediaUrls
     : replyResult.mediaUrl
@@ -1050,6 +1051,7 @@ export async function monitorWebProvider(
       }
 
       const responsePrefix = cfg.messages?.responsePrefix;
+      const textLimit = resolveTextChunkLimit(cfg, "whatsapp");
       let didLogHeartbeatStrip = false;
       let didSendReply = false;
       let toolSendChain: Promise<void> = Promise.resolve();
@@ -1091,6 +1093,7 @@ export async function monitorWebProvider(
               replyResult: toolPayload,
               msg,
               maxMediaBytes,
+              textLimit,
               replyLogger,
               connectionId,
               skipLog: true,
@@ -1134,6 +1137,7 @@ export async function monitorWebProvider(
               replyResult: blockPayload,
               msg,
               maxMediaBytes,
+              textLimit,
               replyLogger,
               connectionId,
               skipLog: true,
@@ -1238,6 +1242,7 @@ export async function monitorWebProvider(
             replyResult: replyPayload,
             msg,
             maxMediaBytes,
+            textLimit,
             replyLogger,
             connectionId,
           });