Send Files and Images with OpenClaw: Complete Guide

Handling files and images in a dev environment can be a pain. You often run into issues with file paths or channel-specific limits that break your workflow. OpenClaw makes this process much easier by managing how media is sent and received across different platforms.

Can OpenClaw send images and PDFs?

Yes. You can put the file in your workspace or provide the full path, then ask the bot to send it. It will pick the right channel method, such as an inline display, an upload, or a hosted link.

Example:

Send /docs/invoice.pdf to the chat

How does OpenClaw handle inbound images?

OpenClaw processes inbound images with vision models like Claude, GPT-4o, and Gemini. The image is passed to the model’s multimodal API if supported.

If vision isn’t available, OpenClaw may describe, OCR, or skip the image depending on your config.

Can skills send images or files?

Yes. Skills can use tool wrappers like fs_write + media_send, or you can just ask the agent to send a file:

Please send the chart at /workspace/output/chart.png

What image formats are supported?

You can use most common formats: PNG, JPEG, GIF, WebP, and HEIC. PDFs are also supported for document-based workflows.

The actual format support depends on the model and channel. Claude and GPT-4o handle most image types; some channels may convert or compress files before the upload is finished.

Is there a file size limit?

Channel-specific limits apply to your files:

• WhatsApp: ~16MB (compressed)
• Telegram: ~50MB for files, ~10MB for photos
• Slack: depends on your workspace tier
• Discord: ~25MB (Nitro: 100MB)

For large files, consider using a hosted link instead.

AI Setup Assistant