switchroom 0.14.16 → 0.14.18

package/dist/agent-scheduler/index.js

@@ -11068,6 +11068,9 @@ var TelegramChannelSchema = exports_external.object({
     short_name: exports_external.string().optional().describe("Telegraph account display name. Defaults to the agent's slug. Used at " + "first-publish to lazily create the account; cached thereafter."),
     author_name: exports_external.string().optional().describe("Telegraph article byline. Defaults to soul.name when set.")
   }).optional().describe("Long-reply publishing via Telegraph (#579). When enabled, replies " + "above the threshold publish as a Telegraph article rendered in " + "Telegram via native Instant View. Off by default — content " + "residency is real for some personas (lawyer, health-coach with PHI). " + "Cascades from defaults.channels.telegram.telegraph. " + "(Migrated from per-agent root in #596.)"),
+  coalesce: exports_external.object({
+    window_ms: exports_external.number().int().nonnegative().optional().describe("Sliding-window (ms) for merging consecutive inbound messages from " + "the same sender+topic into ONE Claude turn. Each new message resets " + "the timer; the turn starts once the sender pauses for this long. " + "Catches forwarded bursts, pasted text the Telegram client split " + "into several messages, and mixed text+media forwards. Default 500. " + "Set 0 to disable (every message becomes its own turn). Raise for " + "users who think in multiple short messages; the trade-off is the " + "single-message turn start is delayed by this much (the \uD83D\uDC40 ack still " + "fires immediately, so perceived latency is unchanged).")
+  }).optional().describe("Inbound coalescing — how the gateway groups rapid consecutive messages " + "into a single turn so a forwarded album or split paste doesn't fan out " + "into N separate turns. Cascades from defaults.channels.telegram.coalesce."),
   webhook_sources: exports_external.array(exports_external.enum(["github", "generic"])).optional().describe("External webhook sources allowed to ingest events into this agent's " + "log. POST /webhook/<agent>/<source> on the switchroom web server. " + "Each source has its own signature verification ('github' = " + "X-Hub-Signature-256 HMAC-SHA256, 'generic' = Bearer token). " + "Per-source secret read from ~/.switchroom/webhook-secrets.json " + "keyed by [agent][source]. Verified events append to " + "<agent>/telegram/webhook-events.jsonl for the agent to read on " + "demand. Off by default — webhook is the only untrusted-inbound " + "surface in the system, so opt-in is mandatory. " + "Cascades from defaults.channels.telegram.webhook_sources. " + "(Migrated from per-agent root in #596 — see #577.)"),
   webhook_dispatch: exports_external.object({
     github: exports_external.array(exports_external.object({

package/dist/auth-broker/index.js

@@ -11068,6 +11068,9 @@ var TelegramChannelSchema = exports_external.object({
     short_name: exports_external.string().optional().describe("Telegraph account display name. Defaults to the agent's slug. Used at " + "first-publish to lazily create the account; cached thereafter."),
     author_name: exports_external.string().optional().describe("Telegraph article byline. Defaults to soul.name when set.")
   }).optional().describe("Long-reply publishing via Telegraph (#579). When enabled, replies " + "above the threshold publish as a Telegraph article rendered in " + "Telegram via native Instant View. Off by default — content " + "residency is real for some personas (lawyer, health-coach with PHI). " + "Cascades from defaults.channels.telegram.telegraph. " + "(Migrated from per-agent root in #596.)"),
+  coalesce: exports_external.object({
+    window_ms: exports_external.number().int().nonnegative().optional().describe("Sliding-window (ms) for merging consecutive inbound messages from " + "the same sender+topic into ONE Claude turn. Each new message resets " + "the timer; the turn starts once the sender pauses for this long. " + "Catches forwarded bursts, pasted text the Telegram client split " + "into several messages, and mixed text+media forwards. Default 500. " + "Set 0 to disable (every message becomes its own turn). Raise for " + "users who think in multiple short messages; the trade-off is the " + "single-message turn start is delayed by this much (the \uD83D\uDC40 ack still " + "fires immediately, so perceived latency is unchanged).")
+  }).optional().describe("Inbound coalescing — how the gateway groups rapid consecutive messages " + "into a single turn so a forwarded album or split paste doesn't fan out " + "into N separate turns. Cascades from defaults.channels.telegram.coalesce."),
   webhook_sources: exports_external.array(exports_external.enum(["github", "generic"])).optional().describe("External webhook sources allowed to ingest events into this agent's " + "log. POST /webhook/<agent>/<source> on the switchroom web server. " + "Each source has its own signature verification ('github' = " + "X-Hub-Signature-256 HMAC-SHA256, 'generic' = Bearer token). " + "Per-source secret read from ~/.switchroom/webhook-secrets.json " + "keyed by [agent][source]. Verified events append to " + "<agent>/telegram/webhook-events.jsonl for the agent to read on " + "demand. Off by default — webhook is the only untrusted-inbound " + "surface in the system, so opt-in is mandatory. " + "Cascades from defaults.channels.telegram.webhook_sources. " + "(Migrated from per-agent root in #596 — see #577.)"),
   webhook_dispatch: exports_external.object({
     github: exports_external.array(exports_external.object({

package/dist/cli/notion-write-pretool.mjs

@@ -11815,6 +11815,9 @@ var TelegramChannelSchema = exports_external.object({
     short_name: exports_external.string().optional().describe("Telegraph account display name. Defaults to the agent's slug. Used at " + "first-publish to lazily create the account; cached thereafter."),
     author_name: exports_external.string().optional().describe("Telegraph article byline. Defaults to soul.name when set.")
   }).optional().describe("Long-reply publishing via Telegraph (#579). When enabled, replies " + "above the threshold publish as a Telegraph article rendered in " + "Telegram via native Instant View. Off by default \u2014 content " + "residency is real for some personas (lawyer, health-coach with PHI). " + "Cascades from defaults.channels.telegram.telegraph. " + "(Migrated from per-agent root in #596.)"),
+  coalesce: exports_external.object({
+    window_ms: exports_external.number().int().nonnegative().optional().describe("Sliding-window (ms) for merging consecutive inbound messages from " + "the same sender+topic into ONE Claude turn. Each new message resets " + "the timer; the turn starts once the sender pauses for this long. " + "Catches forwarded bursts, pasted text the Telegram client split " + "into several messages, and mixed text+media forwards. Default 500. " + "Set 0 to disable (every message becomes its own turn). Raise for " + "users who think in multiple short messages; the trade-off is the " + "single-message turn start is delayed by this much (the \uD83D\uDC40 ack still " + "fires immediately, so perceived latency is unchanged).")
+  }).optional().describe("Inbound coalescing \u2014 how the gateway groups rapid consecutive messages " + "into a single turn so a forwarded album or split paste doesn't fan out " + "into N separate turns. Cascades from defaults.channels.telegram.coalesce."),
   webhook_sources: exports_external.array(exports_external.enum(["github", "generic"])).optional().describe("External webhook sources allowed to ingest events into this agent's " + "log. POST /webhook/<agent>/<source> on the switchroom web server. " + "Each source has its own signature verification ('github' = " + "X-Hub-Signature-256 HMAC-SHA256, 'generic' = Bearer token). " + "Per-source secret read from ~/.switchroom/webhook-secrets.json " + "keyed by [agent][source]. Verified events append to " + "<agent>/telegram/webhook-events.jsonl for the agent to read on " + "demand. Off by default \u2014 webhook is the only untrusted-inbound " + "surface in the system, so opt-in is mandatory. " + "Cascades from defaults.channels.telegram.webhook_sources. " + "(Migrated from per-agent root in #596 \u2014 see #577.)"),
   webhook_dispatch: exports_external.object({
     github: exports_external.array(exports_external.object({

package/dist/cli/switchroom.js

@@ -13632,6 +13632,9 @@ var init_schema = __esm(() => {
       short_name: exports_external.string().optional().describe("Telegraph account display name. Defaults to the agent's slug. Used at " + "first-publish to lazily create the account; cached thereafter."),
       author_name: exports_external.string().optional().describe("Telegraph article byline. Defaults to soul.name when set.")
     }).optional().describe("Long-reply publishing via Telegraph (#579). When enabled, replies " + "above the threshold publish as a Telegraph article rendered in " + "Telegram via native Instant View. Off by default \u2014 content " + "residency is real for some personas (lawyer, health-coach with PHI). " + "Cascades from defaults.channels.telegram.telegraph. " + "(Migrated from per-agent root in #596.)"),
+    coalesce: exports_external.object({
+      window_ms: exports_external.number().int().nonnegative().optional().describe("Sliding-window (ms) for merging consecutive inbound messages from " + "the same sender+topic into ONE Claude turn. Each new message resets " + "the timer; the turn starts once the sender pauses for this long. " + "Catches forwarded bursts, pasted text the Telegram client split " + "into several messages, and mixed text+media forwards. Default 500. " + "Set 0 to disable (every message becomes its own turn). Raise for " + "users who think in multiple short messages; the trade-off is the " + "single-message turn start is delayed by this much (the \uD83D\uDC40 ack still " + "fires immediately, so perceived latency is unchanged).")
+    }).optional().describe("Inbound coalescing \u2014 how the gateway groups rapid consecutive messages " + "into a single turn so a forwarded album or split paste doesn't fan out " + "into N separate turns. Cascades from defaults.channels.telegram.coalesce."),
     webhook_sources: exports_external.array(exports_external.enum(["github", "generic"])).optional().describe("External webhook sources allowed to ingest events into this agent's " + "log. POST /webhook/<agent>/<source> on the switchroom web server. " + "Each source has its own signature verification ('github' = " + "X-Hub-Signature-256 HMAC-SHA256, 'generic' = Bearer token). " + "Per-source secret read from ~/.switchroom/webhook-secrets.json " + "keyed by [agent][source]. Verified events append to " + "<agent>/telegram/webhook-events.jsonl for the agent to read on " + "demand. Off by default \u2014 webhook is the only untrusted-inbound " + "surface in the system, so opt-in is mandatory. " + "Cascades from defaults.channels.telegram.webhook_sources. " + "(Migrated from per-agent root in #596 \u2014 see #577.)"),
     webhook_dispatch: exports_external.object({
       github: exports_external.array(exports_external.object({
@@ -49413,8 +49416,8 @@ var {
 } = import__.default;
 
 // src/build-info.ts
-var VERSION = "0.14.16";
-var COMMIT_SHA = "6f5d9562";
+var VERSION = "0.14.18";
+var COMMIT_SHA = "dddb8617";
 
 // src/cli/agent.ts
 init_source();
@@ -52681,6 +52684,9 @@ function buildAccessJson2(agentConfig, telegramConfig, resolvedTopicId, userId)
   if (tg?.telegraph) {
     access.telegraph = tg.telegraph;
   }
+  if (typeof tg?.coalesce?.window_ms === "number") {
+    access.coalescingGapMs = tg.coalesce.window_ms;
+  }
   return JSON.stringify(access, null, 2) + `
 `;
 }

package/dist/host-control/main.js

@@ -13803,6 +13803,9 @@ var TelegramChannelSchema = exports_external.object({
     short_name: exports_external.string().optional().describe("Telegraph account display name. Defaults to the agent's slug. Used at " + "first-publish to lazily create the account; cached thereafter."),
     author_name: exports_external.string().optional().describe("Telegraph article byline. Defaults to soul.name when set.")
   }).optional().describe("Long-reply publishing via Telegraph (#579). When enabled, replies " + "above the threshold publish as a Telegraph article rendered in " + "Telegram via native Instant View. Off by default — content " + "residency is real for some personas (lawyer, health-coach with PHI). " + "Cascades from defaults.channels.telegram.telegraph. " + "(Migrated from per-agent root in #596.)"),
+  coalesce: exports_external.object({
+    window_ms: exports_external.number().int().nonnegative().optional().describe("Sliding-window (ms) for merging consecutive inbound messages from " + "the same sender+topic into ONE Claude turn. Each new message resets " + "the timer; the turn starts once the sender pauses for this long. " + "Catches forwarded bursts, pasted text the Telegram client split " + "into several messages, and mixed text+media forwards. Default 500. " + "Set 0 to disable (every message becomes its own turn). Raise for " + "users who think in multiple short messages; the trade-off is the " + "single-message turn start is delayed by this much (the \uD83D\uDC40 ack still " + "fires immediately, so perceived latency is unchanged).")
+  }).optional().describe("Inbound coalescing — how the gateway groups rapid consecutive messages " + "into a single turn so a forwarded album or split paste doesn't fan out " + "into N separate turns. Cascades from defaults.channels.telegram.coalesce."),
   webhook_sources: exports_external.array(exports_external.enum(["github", "generic"])).optional().describe("External webhook sources allowed to ingest events into this agent's " + "log. POST /webhook/<agent>/<source> on the switchroom web server. " + "Each source has its own signature verification ('github' = " + "X-Hub-Signature-256 HMAC-SHA256, 'generic' = Bearer token). " + "Per-source secret read from ~/.switchroom/webhook-secrets.json " + "keyed by [agent][source]. Verified events append to " + "<agent>/telegram/webhook-events.jsonl for the agent to read on " + "demand. Off by default — webhook is the only untrusted-inbound " + "surface in the system, so opt-in is mandatory. " + "Cascades from defaults.channels.telegram.webhook_sources. " + "(Migrated from per-agent root in #596 — see #577.)"),
   webhook_dispatch: exports_external.object({
     github: exports_external.array(exports_external.object({

package/dist/vault/approvals/kernel-server.js

@@ -11383,6 +11383,9 @@ var init_schema = __esm(() => {
       short_name: exports_external.string().optional().describe("Telegraph account display name. Defaults to the agent's slug. Used at " + "first-publish to lazily create the account; cached thereafter."),
       author_name: exports_external.string().optional().describe("Telegraph article byline. Defaults to soul.name when set.")
     }).optional().describe("Long-reply publishing via Telegraph (#579). When enabled, replies " + "above the threshold publish as a Telegraph article rendered in " + "Telegram via native Instant View. Off by default — content " + "residency is real for some personas (lawyer, health-coach with PHI). " + "Cascades from defaults.channels.telegram.telegraph. " + "(Migrated from per-agent root in #596.)"),
+    coalesce: exports_external.object({
+      window_ms: exports_external.number().int().nonnegative().optional().describe("Sliding-window (ms) for merging consecutive inbound messages from " + "the same sender+topic into ONE Claude turn. Each new message resets " + "the timer; the turn starts once the sender pauses for this long. " + "Catches forwarded bursts, pasted text the Telegram client split " + "into several messages, and mixed text+media forwards. Default 500. " + "Set 0 to disable (every message becomes its own turn). Raise for " + "users who think in multiple short messages; the trade-off is the " + "single-message turn start is delayed by this much (the \uD83D\uDC40 ack still " + "fires immediately, so perceived latency is unchanged).")
+    }).optional().describe("Inbound coalescing — how the gateway groups rapid consecutive messages " + "into a single turn so a forwarded album or split paste doesn't fan out " + "into N separate turns. Cascades from defaults.channels.telegram.coalesce."),
     webhook_sources: exports_external.array(exports_external.enum(["github", "generic"])).optional().describe("External webhook sources allowed to ingest events into this agent's " + "log. POST /webhook/<agent>/<source> on the switchroom web server. " + "Each source has its own signature verification ('github' = " + "X-Hub-Signature-256 HMAC-SHA256, 'generic' = Bearer token). " + "Per-source secret read from ~/.switchroom/webhook-secrets.json " + "keyed by [agent][source]. Verified events append to " + "<agent>/telegram/webhook-events.jsonl for the agent to read on " + "demand. Off by default — webhook is the only untrusted-inbound " + "surface in the system, so opt-in is mandatory. " + "Cascades from defaults.channels.telegram.webhook_sources. " + "(Migrated from per-agent root in #596 — see #577.)"),
     webhook_dispatch: exports_external.object({
       github: exports_external.array(exports_external.object({

package/dist/vault/broker/server.js

@@ -11383,6 +11383,9 @@ var init_schema = __esm(() => {
       short_name: exports_external.string().optional().describe("Telegraph account display name. Defaults to the agent's slug. Used at " + "first-publish to lazily create the account; cached thereafter."),
       author_name: exports_external.string().optional().describe("Telegraph article byline. Defaults to soul.name when set.")
     }).optional().describe("Long-reply publishing via Telegraph (#579). When enabled, replies " + "above the threshold publish as a Telegraph article rendered in " + "Telegram via native Instant View. Off by default — content " + "residency is real for some personas (lawyer, health-coach with PHI). " + "Cascades from defaults.channels.telegram.telegraph. " + "(Migrated from per-agent root in #596.)"),
+    coalesce: exports_external.object({
+      window_ms: exports_external.number().int().nonnegative().optional().describe("Sliding-window (ms) for merging consecutive inbound messages from " + "the same sender+topic into ONE Claude turn. Each new message resets " + "the timer; the turn starts once the sender pauses for this long. " + "Catches forwarded bursts, pasted text the Telegram client split " + "into several messages, and mixed text+media forwards. Default 500. " + "Set 0 to disable (every message becomes its own turn). Raise for " + "users who think in multiple short messages; the trade-off is the " + "single-message turn start is delayed by this much (the \uD83D\uDC40 ack still " + "fires immediately, so perceived latency is unchanged).")
+    }).optional().describe("Inbound coalescing — how the gateway groups rapid consecutive messages " + "into a single turn so a forwarded album or split paste doesn't fan out " + "into N separate turns. Cascades from defaults.channels.telegram.coalesce."),
     webhook_sources: exports_external.array(exports_external.enum(["github", "generic"])).optional().describe("External webhook sources allowed to ingest events into this agent's " + "log. POST /webhook/<agent>/<source> on the switchroom web server. " + "Each source has its own signature verification ('github' = " + "X-Hub-Signature-256 HMAC-SHA256, 'generic' = Bearer token). " + "Per-source secret read from ~/.switchroom/webhook-secrets.json " + "keyed by [agent][source]. Verified events append to " + "<agent>/telegram/webhook-events.jsonl for the agent to read on " + "demand. Off by default — webhook is the only untrusted-inbound " + "surface in the system, so opt-in is mandatory. " + "Cascades from defaults.channels.telegram.webhook_sources. " + "(Migrated from per-agent root in #596 — see #577.)"),
     webhook_dispatch: exports_external.object({
       github: exports_external.array(exports_external.object({

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "switchroom",
-  "version": "0.14.16",
+  "version": "0.14.18",
   "description": "Run Claude Code 24/7 on your Claude Pro/Max subscription over Telegram. Open-source alternative to OpenClaw and NanoClaw — no API keys.",
   "type": "module",
   "bin": {

package/profiles/_shared/telegram-style.md.hbs

@@ -34,13 +34,14 @@ If both `queued` and `steering` are somehow present, `steering` wins (explicit o
 **Self-narrate the classification.** At the top of your reply for any `steering` or `queued` message, include a brief italic one-liner so the user can correct you — e.g. `_↪️ Treating as steer on the prior task_` or `_📥 Queued as a new task_`.
 
 **Formatting** (Telegram HTML — `reply` and `stream_reply` default to `format: "html"` and convert markdown for you):
-- Use **bold** sparingly for emphasis on key facts only
+- In ordinary one-or-two-line conversational replies, keep **bold** light — emphasis on key facts only, never decoration.
+- **A multi-section message needs visual hierarchy or it reads as a flat wall.** When a reply groups several blocks — a status update, a "where things stand", a message announcing you're dispatching work, a summary with distinct buckets — give each section a **bold label on its own line** and separate sections with **one blank line**. Bold the label (e.g. `**✅ Done / running**`, `**🔲 Remaining**`, `**Next**`); the markdown→HTML converter renders `**label**` as bold, so the eye can find the structure. An emoji prefix with no bold leaves every line at equal weight — that's the plain-text dump to avoid. Bullet lines under a label are fine (`•` or `-`, one point per line).
 - Use `inline code` for filenames, commands, identifiers
 - Use ```fenced code blocks``` for multi-line code
-- Lists are fine; nested lists are not (Telegram flattens them awkwardly)
-- Don't use markdown headings (`##`) in replies — Telegram has no `<h1>` and they render as plain bold lines
-- Keep lines short — long unwrapped lines are hard to read on mobile
-- One idea per message when possible; the user can always ask for more
+- Nested lists are not supported (Telegram flattens them awkwardly) — keep bullets one level deep.
+- Don't use markdown headings (`##`) in replies — bold the label instead (`**Blockers**`, not `## Blockers`).
+- Keep lines short — long unwrapped lines are hard to read on mobile.
+- One idea per message for a quick exchange; a structured update can carry several ideas, but only when each sits under its own bold label with blank-line spacing between them.
 
 **Sound human, not AI.** The canonical list of AI-tells to avoid lives in `SOUL.md` under "Never". Apply those rules to every outbound message, not just long-form. For drafts above ~500 chars, or where you're unsure if the voice lands right, invoke the bundled `/humanizer` skill for a polish pass (it catalogues 29 patterns in detail). If `HUMANIZER_VOICE_FILE` is set and readable, treat its content as the user's personal voice template: match length, tone, vocabulary, and formatting habits described there. The user can generate one with `/humanizer-calibrate`.
 

package/telegram-plugin/dist/gateway/gateway.js

@@ -23737,6 +23737,9 @@ var init_schema = __esm(() => {
       short_name: exports_external.string().optional().describe("Telegraph account display name. Defaults to the agent's slug. Used at " + "first-publish to lazily create the account; cached thereafter."),
       author_name: exports_external.string().optional().describe("Telegraph article byline. Defaults to soul.name when set.")
     }).optional().describe("Long-reply publishing via Telegraph (#579). When enabled, replies " + "above the threshold publish as a Telegraph article rendered in " + "Telegram via native Instant View. Off by default \u2014 content " + "residency is real for some personas (lawyer, health-coach with PHI). " + "Cascades from defaults.channels.telegram.telegraph. " + "(Migrated from per-agent root in #596.)"),
+    coalesce: exports_external.object({
+      window_ms: exports_external.number().int().nonnegative().optional().describe("Sliding-window (ms) for merging consecutive inbound messages from " + "the same sender+topic into ONE Claude turn. Each new message resets " + "the timer; the turn starts once the sender pauses for this long. " + "Catches forwarded bursts, pasted text the Telegram client split " + "into several messages, and mixed text+media forwards. Default 500. " + "Set 0 to disable (every message becomes its own turn). Raise for " + "users who think in multiple short messages; the trade-off is the " + "single-message turn start is delayed by this much (the \uD83D\uDC40 ack still " + "fires immediately, so perceived latency is unchanged).")
+    }).optional().describe("Inbound coalescing \u2014 how the gateway groups rapid consecutive messages " + "into a single turn so a forwarded album or split paste doesn't fan out " + "into N separate turns. Cascades from defaults.channels.telegram.coalesce."),
     webhook_sources: exports_external.array(exports_external.enum(["github", "generic"])).optional().describe("External webhook sources allowed to ingest events into this agent's " + "log. POST /webhook/<agent>/<source> on the switchroom web server. " + "Each source has its own signature verification ('github' = " + "X-Hub-Signature-256 HMAC-SHA256, 'generic' = Bearer token). " + "Per-source secret read from ~/.switchroom/webhook-secrets.json " + "keyed by [agent][source]. Verified events append to " + "<agent>/telegram/webhook-events.jsonl for the agent to read on " + "demand. Off by default \u2014 webhook is the only untrusted-inbound " + "surface in the system, so opt-in is mandatory. " + "Cascades from defaults.channels.telegram.webhook_sources. " + "(Migrated from per-agent root in #596 \u2014 see #577.)"),
     webhook_dispatch: exports_external.object({
       github: exports_external.array(exports_external.object({
@@ -46341,23 +46344,70 @@ function redeliverBufferedInbound(buffer, agent, send, spool) {
   const pending = buffer.drain(agent);
   let redelivered = 0;
   let rebuffered = 0;
-  for (const msg of pending) {
+  for (const { merged, originals } of planBufferedRedelivery(pending)) {
     let delivered = false;
     try {
-      delivered = send(msg);
+      delivered = send(merged);
     } catch {
       delivered = false;
     }
     if (delivered) {
-      redelivered++;
-      spool?.ack(msg);
+      for (const o of originals)
+        spool?.ack(o);
+      redelivered += originals.length;
     } else {
-      buffer.push(agent, msg);
-      rebuffered++;
+      for (const o of originals)
+        buffer.push(agent, o);
+      rebuffered += originals.length;
     }
   }
   return { drained: pending.length, redelivered, rebuffered };
 }
+function isMergeableUserInbound(msg) {
+  return msg.type === "inbound" && (msg.meta == null || msg.meta.source == null);
+}
+function inboundHasMedia(msg) {
+  return msg.imagePath != null || msg.attachment != null;
+}
+function planBufferedRedelivery(pending) {
+  const out = [];
+  let run2 = [];
+  let runHasMedia = false;
+  const sameTarget = (a, b) => a.chatId === b.chatId && (a.threadId ?? null) === (b.threadId ?? null) && a.userId === b.userId;
+  const flush = () => {
+    if (run2.length === 0)
+      return;
+    out.push({ merged: run2.length === 1 ? run2[0] : mergeRun(run2), originals: run2 });
+    run2 = [];
+    runHasMedia = false;
+  };
+  for (const msg of pending) {
+    const msgHasMedia = inboundHasMedia(msg);
+    const canJoin = run2.length > 0 && isMergeableUserInbound(msg) && isMergeableUserInbound(run2[run2.length - 1]) && sameTarget(run2[run2.length - 1], msg) && !(runHasMedia && msgHasMedia);
+    if (!canJoin)
+      flush();
+    run2.push(msg);
+    runHasMedia = runHasMedia || msgHasMedia;
+  }
+  flush();
+  return out;
+}
+function mergeRun(run2) {
+  const last = run2[run2.length - 1];
+  const mediaEntry = run2.find(inboundHasMedia);
+  const merged = {
+    ...last,
+    text: run2.map((m) => m.text).join(`
+`)
+  };
+  delete merged.imagePath;
+  delete merged.attachment;
+  if (mediaEntry?.imagePath != null)
+    merged.imagePath = mediaEntry.imagePath;
+  if (mediaEntry?.attachment != null)
+    merged.attachment = mediaEntry.attachment;
+  return merged;
+}
 function idleDrainTick(buffer, agent, isBridgeAlive, send, spool) {
   if (!agent)
     return null;
@@ -46932,23 +46982,70 @@ function redeliverBufferedInbound2(buffer, agent, send, spool) {
   const pending = buffer.drain(agent);
   let redelivered = 0;
   let rebuffered = 0;
-  for (const msg of pending) {
+  for (const { merged, originals } of planBufferedRedelivery2(pending)) {
     let delivered = false;
     try {
-      delivered = send(msg);
+      delivered = send(merged);
     } catch {
       delivered = false;
     }
     if (delivered) {
-      redelivered++;
-      spool?.ack(msg);
+      for (const o of originals)
+        spool?.ack(o);
+      redelivered += originals.length;
     } else {
-      buffer.push(agent, msg);
-      rebuffered++;
+      for (const o of originals)
+        buffer.push(agent, o);
+      rebuffered += originals.length;
     }
   }
   return { drained: pending.length, redelivered, rebuffered };
 }
+function isMergeableUserInbound2(msg) {
+  return msg.type === "inbound" && (msg.meta == null || msg.meta.source == null);
+}
+function inboundHasMedia2(msg) {
+  return msg.imagePath != null || msg.attachment != null;
+}
+function planBufferedRedelivery2(pending) {
+  const out = [];
+  let run2 = [];
+  let runHasMedia = false;
+  const sameTarget = (a, b) => a.chatId === b.chatId && (a.threadId ?? null) === (b.threadId ?? null) && a.userId === b.userId;
+  const flush = () => {
+    if (run2.length === 0)
+      return;
+    out.push({ merged: run2.length === 1 ? run2[0] : mergeRun2(run2), originals: run2 });
+    run2 = [];
+    runHasMedia = false;
+  };
+  for (const msg of pending) {
+    const msgHasMedia = inboundHasMedia2(msg);
+    const canJoin = run2.length > 0 && isMergeableUserInbound2(msg) && isMergeableUserInbound2(run2[run2.length - 1]) && sameTarget(run2[run2.length - 1], msg) && !(runHasMedia && msgHasMedia);
+    if (!canJoin)
+      flush();
+    run2.push(msg);
+    runHasMedia = runHasMedia || msgHasMedia;
+  }
+  flush();
+  return out;
+}
+function mergeRun2(run2) {
+  const last = run2[run2.length - 1];
+  const mediaEntry = run2.find(inboundHasMedia2);
+  const merged = {
+    ...last,
+    text: run2.map((m) => m.text).join(`
+`)
+  };
+  delete merged.imagePath;
+  delete merged.attachment;
+  if (mediaEntry?.imagePath != null)
+    merged.imagePath = mediaEntry.imagePath;
+  if (mediaEntry?.attachment != null)
+    merged.attachment = mediaEntry.attachment;
+  return merged;
+}
 
 // gateway/inbound-delivery-machine-dispatch.ts
 var enabled6 = process.env.SWITCHROOM_DELIVERY_MACHINE_CUTOVER !== "0";
@@ -51140,10 +51237,10 @@ function sweepStaleTurnActiveMarker(stateDir, opts) {
 }
 
 // ../src/build-info.ts
-var VERSION = "0.14.16";
-var COMMIT_SHA = "6f5d9562";
-var COMMIT_DATE = "2026-05-30T04:37:39Z";
-var LATEST_PR = 2003;
+var VERSION = "0.14.18";
+var COMMIT_SHA = "dddb8617";
+var COMMIT_DATE = "2026-05-30T23:35:26Z";
+var LATEST_PR = 2010;
 var COMMITS_AHEAD_OF_TAG = 0;
 
 // gateway/boot-version.ts
@@ -51605,6 +51702,34 @@ function applySubagentsSchema(db2) {
   }
   db2.exec("CREATE INDEX IF NOT EXISTS subagents_jsonl_id ON subagents(jsonl_agent_id)");
 }
+function mapSubagentRow(row) {
+  return {
+    id: row.id,
+    parent_session_id: row.parent_session_id,
+    parent_turn_key: row.parent_turn_key,
+    agent_type: row.agent_type,
+    description: row.description,
+    background: row.background !== 0,
+    started_at: row.started_at,
+    last_activity_at: row.last_activity_at,
+    ended_at: row.ended_at,
+    status: row.status,
+    result_summary: row.result_summary,
+    jsonl_agent_id: row.jsonl_agent_id
+  };
+}
+function getSubagentByJsonlId(db2, jsonlAgentId) {
+  const row = db2.prepare("SELECT * FROM subagents WHERE jsonl_agent_id = ?").get(jsonlAgentId);
+  return row ? mapSubagentRow(row) : null;
+}
+
+// gateway/worker-feed-dispatch.ts
+function resolveWorkerFeedDispatch(sub, watcherDescription) {
+  return {
+    isBackground: sub?.background ?? false,
+    feedDescription: (sub?.description ?? "") || watcherDescription
+  };
+}
 
 // gateway/resolve-calling-subagent.ts
 function resolveCallingSubagent(opts) {
@@ -52824,19 +52949,22 @@ function looksLikeAuthCode(text) {
     return true;
   return false;
 }
+var bufferedAttachmentKeys = new Set;
 var inboundCoalescer = createInboundCoalescer({
   gapMs: () => loadAccess().coalescingGapMs ?? 500,
   merge: (entries) => {
     const last = entries[entries.length - 1];
+    const withAttachment = entries.find((e) => e.downloadImage != null || e.attachment != null);
     return {
       text: entries.map((e) => e.text).join(`
 `),
       ctx: last.ctx,
-      downloadImage: last.downloadImage,
-      attachment: last.attachment
+      downloadImage: withAttachment?.downloadImage,
+      attachment: withAttachment?.attachment
     };
   },
-  onFlush: (_key, merged) => {
+  onFlush: (key, merged) => {
+    bufferedAttachmentKeys.delete(key);
     handleInbound(merged.ctx, merged.text, merged.downloadImage, merged.attachment);
   }
 });
@@ -55992,19 +56120,29 @@ function safeName(s) {
   return s?.replace(/[<>\[\]\r\n;]/g, "_");
 }
 async function handleInboundCoalesced(ctx, text, downloadImage, attachment) {
-  if (downloadImage || attachment)
-    return handleInbound(ctx, text, downloadImage, attachment);
   if (parseInterruptMarker(text).isInterrupt) {
-    return handleInbound(ctx, text, undefined, undefined);
+    return handleInbound(ctx, text, downloadImage, attachment);
+  }
+  const hasAttachment = downloadImage != null || attachment != null;
+  if (hasAttachment && ctx.message?.media_group_id != null) {
+    return handleInbound(ctx, text, downloadImage, attachment);
   }
   const from = ctx.from;
   if (!from)
     return;
+  if (hasAttachment) {
+    const probeKey = inboundCoalesceKey(String(ctx.chat.id), ctx.message?.message_thread_id, String(from.id));
+    if (bufferedAttachmentKeys.has(probeKey)) {
+      return handleInbound(ctx, text, downloadImage, attachment);
+    }
+  }
   maybeEarlyAckReaction(ctx, from);
   const key = inboundCoalesceKey(String(ctx.chat.id), ctx.message?.message_thread_id, String(from.id));
   const result = inboundCoalescer.enqueue(key, { text, ctx, downloadImage, attachment });
   if (result.bypass)
-    return handleInbound(ctx, text, undefined, undefined);
+    return handleInbound(ctx, text, downloadImage, attachment);
+  if (hasAttachment)
+    bufferedAttachmentKeys.add(key);
 }
 function maybeEarlyAckReaction(ctx, from) {
   const msgId = ctx.message?.message_id;
@@ -60270,7 +60408,7 @@ bot.on("message:text", async (ctx) => {
 });
 bot.on("message:photo", async (ctx) => {
   const caption = ctx.message.caption ?? "(photo)";
-  await handleInbound(ctx, caption, async () => {
+  await handleInboundCoalesced(ctx, caption, async () => {
     const photos = ctx.message.photo;
     const best = photos[photos.length - 1];
     try {
@@ -60306,7 +60444,7 @@ bot.on("message:photo", async (ctx) => {
 bot.on("message:document", async (ctx) => {
   const doc = ctx.message.document;
   const name = safeName(doc.file_name);
-  await handleInbound(ctx, ctx.message.caption ?? `(document: ${name ?? "file"})`, undefined, { kind: "document", file_id: doc.file_id, size: doc.file_size, mime: doc.mime_type, name });
+  await handleInboundCoalesced(ctx, ctx.message.caption ?? `(document: ${name ?? "file"})`, undefined, { kind: "document", file_id: doc.file_id, size: doc.file_size, mime: doc.mime_type, name });
 });
 bot.on("message:voice", async (ctx) => {
   const voice = ctx.message.voice;
@@ -60318,7 +60456,7 @@ bot.on("message:voice", async (ctx) => {
       const text = ctx.message.caption ? `${ctx.message.caption}
 
 [voice transcript] ${transcript}` : `[voice transcript] ${transcript}`;
-      await handleInbound(ctx, text, undefined, {
+      await handleInboundCoalesced(ctx, text, undefined, {
         kind: "voice",
         file_id: voice.file_id,
         size: voice.file_size,
@@ -60327,7 +60465,7 @@ bot.on("message:voice", async (ctx) => {
       return;
     }
   }
-  await handleInbound(ctx, ctx.message.caption ?? "(voice message)", undefined, { kind: "voice", file_id: voice.file_id, size: voice.file_size, mime: voice.mime_type });
+  await handleInboundCoalesced(ctx, ctx.message.caption ?? "(voice message)", undefined, { kind: "voice", file_id: voice.file_id, size: voice.file_size, mime: voice.mime_type });
 });
 async function maybeTranscribeVoice(fileId, mimeType, language) {
   let apiKey = null;
@@ -60387,15 +60525,15 @@ async function maybeTranscribeVoice(fileId, mimeType, language) {
 bot.on("message:audio", async (ctx) => {
   const audio = ctx.message.audio;
   const name = safeName(audio.file_name);
-  await handleInbound(ctx, ctx.message.caption ?? `(audio: ${safeName(audio.title) ?? name ?? "audio"})`, undefined, { kind: "audio", file_id: audio.file_id, size: audio.file_size, mime: audio.mime_type, name });
+  await handleInboundCoalesced(ctx, ctx.message.caption ?? `(audio: ${safeName(audio.title) ?? name ?? "audio"})`, undefined, { kind: "audio", file_id: audio.file_id, size: audio.file_size, mime: audio.mime_type, name });
 });
 bot.on("message:video", async (ctx) => {
   const video = ctx.message.video;
-  await handleInbound(ctx, ctx.message.caption ?? "(video)", undefined, { kind: "video", file_id: video.file_id, size: video.file_size, mime: video.mime_type, name: safeName(video.file_name) });
+  await handleInboundCoalesced(ctx, ctx.message.caption ?? "(video)", undefined, { kind: "video", file_id: video.file_id, size: video.file_size, mime: video.mime_type, name: safeName(video.file_name) });
 });
 bot.on("message:video_note", async (ctx) => {
   const vn = ctx.message.video_note;
-  await handleInbound(ctx, "(video note)", undefined, { kind: "video_note", file_id: vn.file_id, size: vn.file_size });
+  await handleInboundCoalesced(ctx, "(video note)", undefined, { kind: "video_note", file_id: vn.file_id, size: vn.file_size });
 });
 bot.on("message:sticker", async (ctx) => {
   const sticker = ctx.message.sticker;
@@ -60405,13 +60543,13 @@ bot.on("message:sticker", async (ctx) => {
   if (sticker.set_name)
     parts.push(`from "${sticker.set_name}"`);
   const text = parts.length > 0 ? `(sticker \u2014 ${parts.join(" ")})` : "(sticker)";
-  await handleInbound(ctx, text, undefined, { kind: "sticker", file_id: sticker.file_id, size: sticker.file_size });
+  await handleInboundCoalesced(ctx, text, undefined, { kind: "sticker", file_id: sticker.file_id, size: sticker.file_size });
 });
 bot.on("message:animation", async (ctx) => {
   const animation = ctx.message.animation;
   const caption = ctx.message.caption;
   const text = caption ? `(gif) ${caption}` : "(gif)";
-  await handleInbound(ctx, text, undefined, {
+  await handleInboundCoalesced(ctx, text, undefined, {
     kind: "animation",
     file_id: animation.file_id,
     size: animation.file_size,
@@ -61332,8 +61470,6 @@ var didOneTimeSetup = false;
               onFinish: ({ agentId, outcome, description, resultText, toolCount, durationMs }) => {
                 deferredDoneReactions.promote();
                 let fleetChatId = "";
-                let isBackground = false;
-                let dispatchDesc = "";
                 try {
                   const fleets = progressDriver?.peekAllFleets() ?? [];
                   for (const f of fleets) {
@@ -61343,18 +61479,16 @@ var didOneTimeSetup = false;
                     }
                   }
                 } catch {}
+                let dispatch = resolveWorkerFeedDispatch(null, description);
                 if (turnsDb != null) {
                   try {
-                    const row = turnsDb.prepare("SELECT background, description FROM subagents WHERE jsonl_agent_id = ?").get(agentId);
-                    if (row != null) {
-                      isBackground = row.background === 1;
-                      dispatchDesc = row.description ?? "";
-                    }
+                    dispatch = resolveWorkerFeedDispatch(getSubagentByJsonlId(turnsDb, agentId), description);
                   } catch {}
                 }
+                const isBackground = dispatch.isBackground;
                 if (workerFeedEnabled) {
                   workerActivityFeed.finish(agentId, {
-                    description: dispatchDesc || description,
+                    description: dispatch.feedDescription,
                     lastTool: null,
                     toolCount,
                     latestSummary: resultText,
@@ -61396,8 +61530,6 @@ var didOneTimeSetup = false;
               },
               onProgress: ({ agentId, description, latestSummary, elapsedMs, prevBucketIdx, setBucketIdx, lastTool, toolCount }) => {
                 let fleetChatId = "";
-                let isBackground = false;
-                let dispatchDesc = "";
                 try {
                   const fleets = progressDriver?.peekAllFleets() ?? [];
                   for (const f of fleets) {
@@ -61407,20 +61539,18 @@ var didOneTimeSetup = false;
                     }
                   }
                 } catch {}
+                let dispatch = resolveWorkerFeedDispatch(null, description);
                 if (turnsDb != null) {
                   try {
-                    const row = turnsDb.prepare("SELECT background, description FROM subagents WHERE jsonl_agent_id = ?").get(agentId);
-                    if (row != null) {
-                      isBackground = row.background === 1;
-                      dispatchDesc = row.description ?? "";
-                    }
+                    dispatch = resolveWorkerFeedDispatch(getSubagentByJsonlId(turnsDb, agentId), description);
                   } catch {}
                 }
+                const isBackground = dispatch.isBackground;
                 if (!isBackground)
                   return;
                 if (workerFeedEnabled) {
                   workerActivityFeed.update(agentId, fleetChatId || (loadAccess().allowFrom[0] ?? ""), {
-                    description: dispatchDesc || description,
+                    description: dispatch.feedDescription,
                     lastTool,
                     toolCount,
                     latestSummary,