clawgram 2.6.0 → 2.7.0

package/dist/channel.js

@@ -12,6 +12,36 @@ const node_fs_1 = require("node:fs");
  *  a different conversation from a spoken line or a screenshot, and the
  *  transfer is not free. */
 const INBOUND_MEDIA_MAX_BYTES = 25 * 1024 * 1024;
+/**
+ * What this channel promises the Gateway.
+ *
+ * Annotated with core's own `ChannelCapabilities` on purpose: the shape is read
+ * by core (`resolveChannelTtsVoiceDelivery` reaches straight into
+ * `capabilities.tts.voice`), so a typo here would not fail — it would silently
+ * fall back to a default. With the annotation the compiler checks the promise
+ * against the version of OpenClaw we build against.
+ */
+const CHANNEL_CAPABILITIES = {
+    chatTypes: ["direct", "group"],
+    reactions: true,
+    threads: true,
+    media: true,
+    nativeCommands: false,
+    blockStreaming: false,
+    // Without this key core resolves the default "audio-file" and delivers
+    // synthesized speech as a document: a grey file card you must download
+    // before you know what it is. Advertising "voice-note" makes core mark such
+    // sends with `asVoice`, which the upload path honours.
+    //
+    // `transcodesAudio` is deliberately absent: we ship no ffmpeg and add no
+    // dependencies, so core must hand us Ogg/Opus — the only container Telegram
+    // renders as a voice bubble.
+    tts: {
+        voice: {
+            synthesisTarget: "voice-note",
+        },
+    },
+};
 const media_1 = require("./media");
 const channel_runtime_1 = require("openclaw/plugin-sdk/channel-runtime");
 const param_readers_1 = require("openclaw/plugin-sdk/param-readers");
@@ -201,14 +231,7 @@ const createChannelPlugin = (runtimes, pluginRuntime) => {
             blurb: "Connect your personal Telegram account to OpenClaw via MTProto. Your AI assistant responds as you.",
             aliases: ["tguserbot"],
         },
-        capabilities: {
-            chatTypes: ["direct", "group"],
-            reactions: true,
-            threads: true,
-            media: true,
-            nativeCommands: false,
-            blockStreaming: false,
-        },
+        capabilities: CHANNEL_CAPABILITIES,
         agentPrompt: {
             messageToolHints: () => [
                 "Use clawgram to send Telegram replies from the connected personal account.",
@@ -1316,22 +1339,25 @@ const createChannelPlugin = (runtimes, pluginRuntime) => {
                         removed: reactionParams.remove,
                     });
                 }
+                // Core normalizes whichever of these it filled in to a local path (see
+                // `mediaSourceParams` above); `mediaUrl` stays a URL, which GramJS
+                // accepts as well.
+                const attachedFile = (0, param_readers_1.readStringParam)(params, "filePath")
+                    ?? (0, param_readers_1.readStringParam)(params, "path")
+                    ?? (0, param_readers_1.readStringParam)(params, "media")
+                    ?? (0, param_readers_1.readStringParam)(params, "mediaUrl");
                 // Core dispatches `upload-file`; `sendAttachment` is its legacy alias
-                // and arrives from older callers, so both land here.
-                if (action === "upload-file" || action === "sendAttachment") {
+                // and arrives from older callers. A plain `send` carrying a file lands
+                // here too — `openclaw message send --media` does exactly that, and
+                // routing it to the text path dropped the file without a word.
+                if (action === "upload-file" || action === "sendAttachment" || (action === "send" && attachedFile)) {
                     const rawUploadTo = (0, helpers_1.resolveActionTarget)(params, toolContext);
                     const uploadTo = (0, helpers_1.normalizeOutboundTarget)(rawUploadTo);
                     const uploadAccountId = resolveRuntimeAccountId(cfg, accountId);
                     if (!uploadAccountId) {
                         throw new Error("clawgram: no configured account found");
                     }
-                    // Core normalizes whichever of these it filled in to a local path
-                    // (see `mediaSourceParams` above); `mediaUrl` stays a URL, which
-                    // GramJS accepts as well.
-                    const file = (0, param_readers_1.readStringParam)(params, "filePath")
-                        ?? (0, param_readers_1.readStringParam)(params, "path")
-                        ?? (0, param_readers_1.readStringParam)(params, "media")
-                        ?? (0, param_readers_1.readStringParam)(params, "mediaUrl");
+                    const file = attachedFile;
                     if (!file) {
                         throw new Error("clawgram: upload-file requires filePath, path, media, or mediaUrl");
                     }
@@ -1343,6 +1369,7 @@ const createChannelPlugin = (runtimes, pluginRuntime) => {
                         : captionText.replaceAll("\\n", "\n");
                     const uploadReplyToId = (0, param_readers_1.readStringOrNumberParam)(params, "replyToId") ?? (0, param_readers_1.readStringOrNumberParam)(params, "replyTo");
                     const uploadThreadId = (0, param_readers_1.readStringOrNumberParam)(params, "threadId");
+                    const asVoice = (0, helpers_1.readVoiceNoteFlag)(params);
                     actionLog.info("clawgram handleAction upload-file", {
                         accountId: uploadAccountId,
                         dryRun: dryRun === true,
@@ -1350,6 +1377,7 @@ const createChannelPlugin = (runtimes, pluginRuntime) => {
                         hasCaption: Boolean(caption),
                         replyToId: uploadReplyToId ?? null,
                         threadId: uploadThreadId ?? null,
+                        asVoice,
                     });
                     if (dryRun === true) {
                         return (0, core_1.jsonResult)({
@@ -1369,6 +1397,7 @@ const createChannelPlugin = (runtimes, pluginRuntime) => {
                         caption: caption || undefined,
                         replyToMessageId: (0, helpers_1.resolveReplyToMessageIdForTarget)(rawUploadTo, uploadReplyToId),
                         messageThreadId: parseOptionalThreadId(uploadThreadId),
+                        asVoice,
                     });
                     actionLog.info("clawgram handleAction upload-file completed", {
                         accountId: uploadAccountId,

package/dist/gramjs-client.js

@@ -1,6 +1,7 @@
 "use strict";
 Object.defineProperty(exports, "__esModule", { value: true });
 exports.GramJsClientManager = void 0;
+exports.buildVoiceNoteParams = buildVoiceNoteParams;
 const telegram_1 = require("telegram");
 const sessions_1 = require("telegram/sessions");
 const helpers_1 = require("./helpers");
@@ -79,6 +80,22 @@ function parseTargetWithThread(rawTarget) {
         chatId: raw,
     };
 }
+/**
+ * Voice-message option for `sendFile`.
+ *
+ * GramJS branches on `voiceNote` and builds `DocumentAttributeAudio` with
+ * `voice: true` itself, so the attribute must not be assembled by hand.
+ * The key is omitted rather than set to `false`: this codebase already paid
+ * for the lesson that a key's mere presence can flip a branch (see the MTProxy
+ * note in `proxy-config.ts`).
+ *
+ * Telegram renders a voice bubble only for Ogg/Opus; core is responsible for
+ * handing us that container, which is why the channel does not advertise
+ * `transcodesAudio`.
+ */
+function buildVoiceNoteParams(asVoice) {
+    return asVoice === true ? { voiceNote: true } : {};
+}
 function buildForumReplyParams(messageThreadId, replyToMessageId) {
     const normalizedThreadId = typeof messageThreadId === "number" && Number.isFinite(messageThreadId)
         ? Math.trunc(messageThreadId)
@@ -502,6 +519,7 @@ class GramJsClientManager {
             file: args.file,
             caption: args.caption,
             ...replyParams,
+            ...buildVoiceNoteParams(args.asVoice),
         });
     }
 }

package/dist/helpers.js

@@ -15,6 +15,7 @@ exports.readLatestAssistantFallbackFromTranscript = readLatestAssistantFallbackF
 exports.resolveActionTarget = resolveActionTarget;
 exports.resolveReplyToMessageIdForTarget = resolveReplyToMessageIdForTarget;
 exports.readMessageText = readMessageText;
+exports.readVoiceNoteFlag = readVoiceNoteFlag;
 exports.resolveAllowFrom = resolveAllowFrom;
 exports.resolveGroupPolicy = resolveGroupPolicy;
 exports.resolveGroups = resolveGroups;
@@ -193,6 +194,17 @@ function resolveReplyToMessageIdForTarget(rawTarget, replyToId) {
     }
     return undefined;
 }
+/**
+ * Whether an outbound file should become a Telegram voice message.
+ *
+ * Core emits `asVoice` and, on some paths, the older `audioAsVoice` — both
+ * carry the same meaning, so both are read. Only a real `true` counts: a
+ * string "true" or a stray truthy value must not silently turn a document
+ * into a voice bubble.
+ */
+function readVoiceNoteFlag(params) {
+    return params?.asVoice === true || params?.audioAsVoice === true;
+}
 function readMessageText(params) {
     const message = (0, core_1.readStringParam)(params, "message", { allowEmpty: true });
     if (typeof message === "string") {

package/openclaw.plugin.json

@@ -2,7 +2,7 @@
   "id": "clawgram",
   "name": "Clawgram",
   "description": "Clawgram — personal Telegram (MTProto userbot) channel for OpenClaw. Your AI assistant reads and responds as you.",
-  "version": "2.6.0",
+  "version": "2.7.0",
   "configSchema": {
     "type": "object",
     "additionalProperties": false,

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "clawgram",
-  "version": "2.6.0",
+  "version": "2.7.0",
   "description": "Clawgram — personal Telegram (MTProto userbot) channel for OpenClaw. Your AI assistant reads and responds as you.",
   "main": "./dist/index.js",
   "scripts": {