Hermes Agent does not respond to uploading files after connecting to Open WebUI, so don't toss the front-end first. The official API Server documentation has been written quite bluntly: file uploads via the API are not currently supported. In other words, even if there is an upload button on the front-end, the Hermes Agent back-end cannot process the uploaded documents or images as API file inputs.
What does this mean?
If you want to do document analysis, image understanding, and attachment Q&A, the most reliable way is not to continue to change the Open WebUI, but to change the path:
- Place the file in a local directory that Hermes Agent can access, and let it read it using the file/terminal tool.
- Or use the CLI / gateway scenario instead, and enter the session with the platform's native attachment capability.
- Or extract the contents of the file into text and send a text message through the API.
When is not a problem
If you upload attachments in Telegram, Discord, Slack, Email, and other gateways, it means that Hermes itself is not broken, and the limitations are only at the API Server level. Don't think of "gateway support file" and "OpenAI compatible API support file" as the same thing.
So the answer to this question is actually very short: it's not that the Open WebUI is mismatched, but that the Hermes Agent API server hasn't uploaded the file yet. In the short term, going to the local file path or text extraction is more time-saving than continuing to cut the front-end upload button.
Official open source address: https://github.com/NousResearch/hermes-agent; Official document entry: https://hermes-agent.nousresearch.com/.
