Error 429 and HTTP Request tool for AI Agent

Questions
1

Hey there !

I’m trying to build a simple version of deep search, but I meet an error 429 since the AI Agent node seems to make all the request almost at the same time.

However, since there’s no batching or retry on error on the HTTP Request tool (as opposed to the HTTP request node), I don’t know what to do about it.

Thank you for your help !

Describe the problem/error/question

When using the AI Agent node, it seems to call all the HTTP requests almost at the same time. This triggers a 429 error.

What is the error message (if any)?

"There was an error: “Request failed with status code 429"”

Please share your workflow

Information on your n8n setup

  • n8nVersion: 1.77.3
  • platform: docker (self-hosted)
  • nodeJsVersion: 20.18.2
  • database: postgres
  • executionMode: scaling
2

It looks like your topic is missing some important information. Could you provide the following if applicable.

  • n8n version:
  • Database (default: SQLite):
  • n8n EXECUTIONS_PROCESS setting (default: own, main):
  • Running n8n via (Docker, npm, n8n cloud, desktop app):
  • Operating system:
3

Hey @AntoineDsh

Can you try adding a wait node before sending it to the Agent?

4

Tried it, it does nothing. Also tried with loop over item, but it won’t change anything since it’s the agent itself that call 3-5 times the HTTP subnode

1 Like
5

Instead of using HTTP Tool Node, Why not move the HTTP Request to a different workflow and use it as a Custom tool

Here is my main agent workflow

Here is my custom tool workflow

6

Hi mcnaveen,
Thank you for your responses.

I tried that as well, but unfortunately, the workflow is being triggered multiple times simultaneously, causing the same error.

7

It should work. Which API endpoint are you calling?

RapidAPI?

8

Yes this is the API : Google API

9

I am facing a similar problem here

Captura de tela 2025-02-12 100112
Captura de tela 2025-02-12 1001121142×926 53 KB

10

Hey @Jon Need your inputs

11

Hey :slight_smile: Did you find any workaround for this ?

12

Hey @AntoineDsh

You can use the AI Agent to prepare the correct output and then use the standard HTTP request node AFTER the agent in the workflow to set up any retries and failures.
Have you tried that approach?

13

This topic was automatically closed 90 days after the last reply. New replies are no longer allowed.