@friendlyrobot/discord-pi-agent 0.13.0 → 0.14.0

package/dist/index.js

@@ -870,15 +870,15 @@ var logger5 = createModuleLogger("image-description");
 async function describeImage(agentService, imageData, mimeType, userText, visionModel) {
   const session = await agentService.createTemporarySession();
   await session.setModel(visionModel);
-  const isPdf = mimeType === "application/pdf";
+  const mediaType = getMediaType(mimeType);
   const imageContent = {
     type: "image",
     data: imageData,
     mimeType
   };
   let promptText;
-  if (isPdf) {
-    promptText = userText.trim().length > 0 ? `The user sent a PDF document with the following message: "${userText}". Please extract and summarize the text content of this PDF. Be thorough — include all important details, sections, and data from the document.` : "Please extract and summarize the text content of this PDF document. Be thorough — include all important details, sections, data, and key points.";
+  if (mediaType === "document") {
+    promptText = userText.trim().length > 0 ? `The user sent a document with the following message: "${userText}". Please extract and summarize the text content of this document. Be thorough — include all important details, sections, and data from the document.` : "Please extract and summarize the text content of this document. Be thorough — include all important details, sections, data, and key points.";
   } else {
     promptText = userText.trim().length > 0 ? `The user sent this image with the following message: "${userText}". Please describe the image in detail and address any questions from the user's message.` : "Please describe this image in detail. What do you see?";
   }
@@ -920,6 +920,12 @@ function extractLastAssistantText(session) {
   }
   return "";
 }
+function getMediaType(mimeType) {
+  if (mimeType.startsWith("image/")) {
+    return "image";
+  }
+  return "document";
+}
 function isAssistantMessage(msg) {
   return typeof msg === "object" && msg !== null && "role" in msg && msg.role === "assistant";
 }
@@ -1223,9 +1229,24 @@ var MEDIA_ATTACHMENT_EXTENSIONS = [
   ".jpeg",
   ".gif",
   ".webp",
-  ".pdf"
+  ".pdf",
+  ".docx",
+  ".doc",
+  ".pptx",
+  ".ppt",
+  ".xlsx",
+  ".xls"
 ];
 var MAX_MEDIA_ATTACHMENT_SIZE = 25 * 1024 * 1024;
+var OFFICE_MIME_TYPES = new Set([
+  "application/pdf",
+  "application/vnd.openxmlformats-officedocument.wordprocessingml.document",
+  "application/msword",
+  "application/vnd.openxmlformats-officedocument.presentationml.presentation",
+  "application/vnd.ms-powerpoint",
+  "application/vnd.openxmlformats-officedocument.spreadsheetml.sheet",
+  "application/vnd.ms-excel"
+]);
 function isMediaAttachment(attachment) {
   const ext = attachment.name?.slice(attachment.name.lastIndexOf(".")).toLowerCase();
   if (!ext || !MEDIA_ATTACHMENT_EXTENSIONS.includes(ext)) {
@@ -1235,7 +1256,7 @@ function isMediaAttachment(attachment) {
   if (!ct) {
     return false;
   }
-  return ct.startsWith("image/") || ct === "application/pdf";
+  return ct.startsWith("image/") || OFFICE_MIME_TYPES.has(ct);
 }
 async function readMediaAttachments(message) {
   const attachments = message.attachments;
@@ -1297,6 +1318,21 @@ function parseVisionModelId(visionModelId) {
     modelId: trimmed.substring(slashIndex + 1)
   };
 }
+function getMediaLabel(filename, mimeType) {
+  if (mimeType === "application/pdf") {
+    return `[PDF: ${filename}]`;
+  }
+  if (mimeType === "application/vnd.openxmlformats-officedocument.wordprocessingml.document" || mimeType === "application/msword") {
+    return `[Word: ${filename}]`;
+  }
+  if (mimeType === "application/vnd.openxmlformats-officedocument.spreadsheetml.sheet" || mimeType === "application/vnd.ms-excel") {
+    return `[Excel: ${filename}]`;
+  }
+  if (mimeType === "application/vnd.openxmlformats-officedocument.presentationml.presentation" || mimeType === "application/vnd.ms-powerpoint") {
+    return `[PowerPoint: ${filename}]`;
+  }
+  return `[Image: ${filename}]`;
+}
 async function resolveMediaAttachments(media, content, currentModel, config, agentService) {
   const modelSupportsVision = currentModel?.input.includes("image") ?? false;
   if (modelSupportsVision) {
@@ -1319,7 +1355,7 @@ async function resolveMediaAttachments(media, content, currentModel, config, age
     const note = `
 
 [User sent media attachment(s): ${names}]
-` + "(Media vision not configured. Set visionModelId to enable image/PDF understanding.)";
+` + "(Media vision not configured. Set visionModelId to enable image/PDF/document understanding.)";
     return { content: content ? content + note : note, images: [] };
   }
   const parsed = parseVisionModelId(config.visionModelId);
@@ -1342,9 +1378,8 @@ async function resolveMediaAttachments(media, content, currentModel, config, age
   }, "describing media with vision model");
   const descriptions = [];
   for (const m of media) {
-    const isPdf = m.mimeType === "application/pdf";
     const description = await describeImage(agentService, m.data, m.mimeType, content, visionModel);
-    const label = isPdf ? `[PDF: ${m.filename}]` : `[Image: ${m.filename}]`;
+    const label = getMediaLabel(m.filename, m.mimeType);
     descriptions.push(`${label}
 ${description}`);
   }

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@friendlyrobot/discord-pi-agent",
-  "version": "0.13.0",
+  "version": "0.14.0",
   "description": "Reusable Discord gateway bridge for persistent pi agent sessions",
   "license": "MIT",
   "type": "module",