File Attachments
You can attach documents and images to any message. Kognar reads the content and makes it available to the AI model as part of your conversation.
How to attach files
There are two ways to attach files:
- Click the paperclip icon in the input bar
- Drag and drop a file directly onto the chat window
- Press
Ctrl/Cmd+Oto open a file picker
Attached files appear as pills above the input field. Click the × on any pill to remove a file before sending.
Supported file types
Documents
The following document types are read as text and inserted into the conversation context:
| Extension | Description |
|---|---|
.txt | Plain text |
.md | Markdown |
.pdf | PDF documents |
.docx | Microsoft Word |
.xlsx | Microsoft Excel |
.csv | Comma-separated values |
.json | JSON data |
.xml | XML data |
Images
Images are sent as visual input to vision-capable models:
| Extension | Description |
|---|---|
.jpg / .jpeg | JPEG images |
.png | PNG images |
.gif | GIF (static or animated) |
.webp | WebP images |
Size limits
| Limit | Value |
|---|---|
| Per image | 5 MB maximum |
| Total images per message | 25 MB maximum |
| Documents | No explicit limit (practical limit ~100 MB) |
If a file exceeds the limit, an error message appears and the file is not attached.
How content is sent
- Documents are read and their text content is included in your message as additional context.
- Images are encoded and sent alongside your message to vision-capable models. If the selected model doesn't support vision, image attachments are ignored.
Attach a spreadsheet or CSV and ask Kognar to analyze it, find patterns, or generate a chart. Combine with the Analyst agent for best results.
Viewing attached images
Attached images are previewed in your message after sending. The AI model can describe, analyze, or reference visual content in the image.