feat(gateway): implement multimodal (image/document/audio) inbound support for LINE and Telegram

docs/line.md

@@ -83,6 +83,8 @@ In the LINE Developers Console → **Messaging API** tab → scan the QR code wi
 
 - **1:1 chat** — send a message to the bot, get an AI agent response
 - **Group chat** — add the bot to a group, it responds to all messages
+- **Images** — send image messages to the bot (automatically compressed and resized)
+- **Audio** — send audio messages (e.g. voice notes). They are automatically transcribed (if STT is enabled in Core) and passed to the agent as text.
 - **Webhook signature validation** — HMAC-SHA256 via `LINE_CHANNEL_SECRET`
 
 ### Not Supported (LINE API limitations)

docs/telegram.md

@@ -168,6 +168,12 @@ explain VPC peering              ← ignored in groups
 
 DMs and replies within forum topics always trigger the agent (no @mention needed).
 
+### File Attachments
+
+- **Images** — send photos (compressed/resized automatically).
+- **Documents** — send text-based files (e.g. `.txt`, `.csv`, `.rs`, `.py`) up to 512KB. They are passed directly to the agent as text.
+- **Audio/Voice** — send voice notes or audio files. They are automatically transcribed (if STT is enabled in Core) and passed to the agent as text.
+
 ### Emoji reactions
 
 The bot shows status reactions on your message as the agent works:

gateway/src/adapters/feishu.rs

@@ -1,3 +1,4 @@
+use crate::media::{resize_and_compress, FILE_MAX_DOWNLOAD, IMAGE_MAX_DOWNLOAD};
 use crate::schema::*;
 use axum::extract::State;
 use prost::Message as ProstMessage;
@@ -1268,38 +1269,6 @@ pub enum MediaRef {
     File { message_id: String, file_key: String, file_name: String },
 }
 
-const IMAGE_MAX_DIMENSION_PX: u32 = 1200;
-const IMAGE_JPEG_QUALITY: u8 = 75;
-const IMAGE_MAX_DOWNLOAD: u64 = 10 * 1024 * 1024; // 10 MB
-const FILE_MAX_DOWNLOAD: u64 = 512 * 1024; // 512 KB
-
-/// Resize image so longest side <= 1200px, then encode as JPEG.
-/// GIFs are passed through unchanged to preserve animation.
-fn resize_and_compress(raw: &[u8]) -> Result<(Vec<u8>, String), image::ImageError> {
-    use image::ImageReader;
-    use std::io::Cursor;
-
-    let reader = ImageReader::new(Cursor::new(raw)).with_guessed_format()?;
-    let format = reader.format();
-    if format == Some(image::ImageFormat::Gif) {
-        return Ok((raw.to_vec(), "image/gif".to_string()));
-    }
-    let img = reader.decode()?;
-    let (w, h) = (img.width(), img.height());
-    let img = if w > IMAGE_MAX_DIMENSION_PX || h > IMAGE_MAX_DIMENSION_PX {
-        let max_side = std::cmp::max(w, h);
-        let ratio = f64::from(IMAGE_MAX_DIMENSION_PX) / f64::from(max_side);
-        let new_w = (f64::from(w) * ratio) as u32;
-        let new_h = (f64::from(h) * ratio) as u32;
-        img.resize(new_w, new_h, image::imageops::FilterType::Lanczos3)
-    } else {
-        img
-    };
-    let mut buf = Cursor::new(Vec::new());
-    let encoder = image::codecs::jpeg::JpegEncoder::new_with_quality(&mut buf, IMAGE_JPEG_QUALITY);
-    img.write_with_encoder(encoder)?;
-    Ok((buf.into_inner(), "image/jpeg".to_string()))
-}
 
 /// Download a Feishu image by message_id + image_key → resize/compress → base64 Attachment.
 pub async fn download_feishu_image(

gateway/src/adapters/googlechat.rs

@@ -1371,6 +1371,7 @@ mod tests {
             content: Content {
                 content_type: "text".into(),
                 text: "hello".into(),
+                attachments: Vec::new(),
             },
             command: None,
             request_id: Some("req_123".into()),
@@ -1413,6 +1414,7 @@ mod tests {
             content: Content {
                 content_type: "text".into(),
                 text: "hello".into(),
+                attachments: Vec::new(),
             },
             command: None,
             request_id: Some("req_fail".into()),
@@ -1459,6 +1461,7 @@ mod tests {
             content: Content {
                 content_type: "text".into(),
                 text: "".into(),
+                attachments: Vec::new(),
             },
             command: None,
             request_id: Some("req_empty".into()),
@@ -1502,6 +1505,7 @@ mod tests {
             content: Content {
                 content_type: "text".into(),
                 text: long_text,
+                attachments: Vec::new(),
             },
             command: None,
             request_id: Some("req_multi_fail".into()),
@@ -1535,6 +1539,7 @@ mod tests {
             content: Content {
                 content_type: "text".into(),
                 text: "hello".into(),
+                attachments: Vec::new(),
             },
             command: None,
             request_id: Some("req_notoken".into()),
@@ -1579,6 +1584,7 @@ mod tests {
             content: Content {
                 content_type: "text".into(),
                 text: "updated text".into(),
+                attachments: Vec::new(),
             },
             command: Some("edit_message".into()),
             request_id: None,
@@ -1620,6 +1626,7 @@ mod tests {
             content: Content {
                 content_type: "text".into(),
                 text: long_text,
+                attachments: Vec::new(),
             },
             command: None,
             request_id: Some("req_multi".into()),
@@ -1676,6 +1683,7 @@ mod tests {
             content: Content {
                 content_type: "text".into(),
                 text: long_text,
+                attachments: Vec::new(),
             },
             command: None,
             request_id: Some("req_partial".into()),

gateway/src/adapters/line.rs

@@ -1,3 +1,4 @@
+use crate::media::{resize_and_compress, AUDIO_MAX_DOWNLOAD, IMAGE_MAX_DOWNLOAD};
 use crate::schema::*;
 use axum::extract::State;
 use serde::Deserialize;
@@ -90,45 +91,55 @@ pub async fn webhook(
         let Some(ref msg) = event.message else {
             continue;
         };
-        if msg.message_type != "text" {
+        let is_text = msg.message_type == "text";
+        let is_image = msg.message_type == "image";
+        let is_audio = msg.message_type == "audio";
+
+        if !is_text && !is_image && !is_audio {
             continue;
         }
-        let Some(ref text) = msg.text else {
-            continue;
-        };
-        if text.trim().is_empty() {
+
+        let text = msg.text.clone().unwrap_or_default();
+        if is_text && text.trim().is_empty() {
             continue;
         }
 
+        let mut attachments = Vec::new();
+        if is_image || is_audio {
+            if let Some(ref access_token) = state.line_access_token {
+                let client = &state.client;
+                let att_type = if is_image { "image" } else { "audio" };
+                if let Some(att) = download_line_media(client, access_token, &msg.id, att_type).await {
+                    attachments.push(att);
+                }
+            } else {
+                warn!("LINE media received but LINE_CHANNEL_ACCESS_TOKEN not set");
+            }
+        }
+
         let source = event.source.as_ref();
         let (channel_id, channel_type) = match source {
-            Some(s) if s.source_type == "group" => {
-                match s.group_id.as_deref() {
-                    Some(id) if !id.is_empty() => (id.to_string(), "group".to_string()),
-                    _ => {
-                        warn!("LINE group event missing groupId, skipping");
-                        continue;
-                    }
+            Some(s) if s.source_type == "group" => match s.group_id.as_deref() {
+                Some(id) if !id.is_empty() => (id.to_string(), "group".to_string()),
+                _ => {
+                    warn!("LINE group event missing groupId, skipping");
+                    continue;
                 }
-            }
-            Some(s) if s.source_type == "room" => {
-                match s.room_id.as_deref() {
-                    Some(id) if !id.is_empty() => (id.to_string(), "room".to_string()),
-                    _ => {
-                        warn!("LINE room event missing roomId, skipping");
-                        continue;
-                    }
+            },
+            Some(s) if s.source_type == "room" => match s.room_id.as_deref() {
+                Some(id) if !id.is_empty() => (id.to_string(), "room".to_string()),
+                _ => {
+                    warn!("LINE room event missing roomId, skipping");
+                    continue;
                 }
-            }
-            Some(s) => {
-                match s.user_id.as_deref() {
-                    Some(id) if !id.is_empty() => (id.to_string(), "user".to_string()),
-                    _ => {
-                        warn!("LINE user event missing userId, skipping");
-                        continue;
-                    }
+            },
+            Some(s) => match s.user_id.as_deref() {
+                Some(id) if !id.is_empty() => (id.to_string(), "user".to_string()),
+                _ => {
+                    warn!("LINE user event missing userId, skipping");
+                    continue;
                 }
-            }
+            },
             None => {
                 warn!("LINE event missing source, skipping");
                 continue;
@@ -138,7 +149,7 @@ pub async fn webhook(
             .and_then(|s| s.user_id.as_deref())
             .unwrap_or("unknown");
 
-        let gateway_event = GatewayEvent::new(
+        let mut gateway_event = GatewayEvent::new(
             "line",
             ChannelInfo {
                 id: channel_id.clone(),
@@ -151,10 +162,11 @@ pub async fn webhook(
                 display_name: user_id.into(),
                 is_bot: false,
             },
-            text,
+            &text,
             &msg.id,
             vec![],
         );
+        gateway_event.content.attachments = attachments;
 
         // Cache the reply token for hybrid Reply/Push dispatch
         if let Some(ref reply_token) = event.reply_token {
@@ -266,3 +278,81 @@ pub async fn dispatch_line_reply(
 
     used_reply
 }
+
+/// Download media content from LINE Messaging API.
+async fn download_line_media(
+    client: &reqwest::Client,
+    access_token: &str,
+    message_id: &str,
+    attachment_type: &str,
+) -> Option<Attachment> {
+    let url = format!(
+        "https://api-data.line.me/v2/bot/message/{}/content",
+        message_id
+    );
+    let resp = client
+        .get(url)
+        .bearer_auth(access_token)
+        .send()
+        .await
+        .ok()?;
+
+    if !resp.status().is_success() {
+        error!(status = %resp.status(), "LINE media download failed");
+        return None;
+    }
+
+    let max_size = if attachment_type == "image" {
+        IMAGE_MAX_DOWNLOAD
+    } else {
+        AUDIO_MAX_DOWNLOAD
+    };
+
+    if let Some(cl) = resp.headers().get(reqwest::header::CONTENT_LENGTH) {
+        if let Ok(size) = cl.to_str().unwrap_or("0").parse::<u64>() {
+            if size > max_size {
+                warn!(message_id, size, "LINE {} Content-Length exceeds limit", attachment_type);
+                return None;
+            }
+        }
+    }
+
+    let content_type = resp
+        .headers()
+        .get(reqwest::header::CONTENT_TYPE)
+        .and_then(|h| h.to_str().ok())
+        .unwrap_or(if attachment_type == "image" { "image/jpeg" } else { "audio/mp4" })
+        .to_string();
+
+    let bytes = resp.bytes().await.ok()?;
+    if bytes.len() as u64 > max_size {
+        warn!(message_id, size = bytes.len(), "LINE {} exceeds limit", attachment_type);
+        return None;
+    }
+
+    let (data_bytes, mime, filename) = if attachment_type == "image" {
+        match resize_and_compress(&bytes) {
+            Ok((c, _m)) => (c, content_type, format!("{}.jpg", message_id)),
+            Err(e) => {
+                error!(err = %e, "LINE image processing failed");
+                return None;
+            }
+        }
+    } else {
+        // For audio, we don't process, just send as is.
+        // LINE audio is usually m4a.
+        (bytes.to_vec(), content_type, format!("{}.m4a", message_id))
+    };
+
+    use base64::Engine;
+    let b64_data = base64::engine::general_purpose::STANDARD.encode(&data_bytes);
+    info!(message_id, size = data_bytes.len(), "LINE {} download successful", attachment_type);
+
+    Some(Attachment {
+        attachment_type: attachment_type.into(),
+        filename,
+        mime_type: mime,
+        data: b64_data,
+        size: data_bytes.len() as u64,
+    })
+}