Hi all!
I have tried to create the workflow with chat + AI agent + memory + HTTP Request node which send files to my custom server.
The version of the workflow is last: 1.89(Beta)
I have watched several video and read the documentation in which i found that there is ability to send files via HTTP Request node using ‘turn on binary support’.
In fact, my configuration of HTTP request has difference. There is not any button which turn on the file support mentioned in documentation and video.
Please help me to turn on the file support in the HTTP Request. I will attach the screen shot from video where I saw such support and my screenshot on which this file support does not exist.
Waiting for reply anybody who has understanding, in what I was wrong and how to fix it.
Hi, could you fix it? I have the same issue. I think agents cant pass binary items, when I use it without Agents (Just pass the binary from chat to http request) I can.
Hello Dmytro Tymoshenko
Please send your workflow or a photo of the workflow.
@Dmytro_Tymoshenko @Homero_Monsalvo
Could this be the downstream you’re thinking of?
Hi @Dmytro_Tymoshenko, your first screenshot has the AI Agent connected to the tools with HTTP node. The second screenshot is just the HTTP node that is not connected to the AI Agent node.
Hi,
Yes you are right, but the HTTP Request node which is connected to AI agent has different functionality. It does not have ability to end files, which is very useful for me in our business case. I was surprised that the version of HTTP request could be different inside one particular workflow. Do you know, could I config the flow to use the newest version of HTTP Request node in composition with AI Agent? Because for me, the file processing is that business case, which I need to cover my target flow. Thanks in advance.
Hi,
I will check your flow, and answer later today, is it work or not.
My current flow with which I came here is next:
I will be very happy to get the description or steps how to implement such behaviour like I shown in the image if the flow you mentioned above will not work (i will check it and return back). The target is to send one or several files in the chat, by AI model figure out that it should be sent to the tool or server in one request and receive one file from the server to the chat or (if it impossible to receive file, receive link to it to download)
Thank you in advance.
Dmytro
WORKFLOW | ERRORS | FILE Processing
Thank you @interss for your help and created JSON with workflow.
I have got provided JSON of workflow from you and fixed on my side (because it was not working as is)
But I bumped into issues to get binary from the AI Agent context.
@interss or @Homero_Monsalvo or @waynemorphic
Please, could you check my workflow and maybe make any changes in it. I tried several variants of mapping but I stucked and dont know how to fix it…
I added the JSON below with the workflow, maybe anybody could take a look and help me to fix configs. I am a newcomer in the N8N and maybe I dont understand something. In addition I tried to get solutions from internal N8N AI assistant but generated code and solutions from internal AI does not work.
[quote=“Uftiman, post:10, topic:104939”]
What should you do?
The file (binary data) should be handled by a separate path, bypassing the AI Agent.
The AI Agent should only return commands or metadata, which can then be processed by the File Processor.
Advice:
If you want to send files, send them directly to the File Processor or HTTP Request, not through the AI Agent.
Use an IF node to separate the logic:
if there’s a file — use one branch, if it’s text — use another.
