AI Agent Tool (v3) chamando tools em duplicidade após atualização n8n

Questions
1

AI Agent Tool (v3) executing workflow tools twice after n8n update

After updating n8n yesterday, my workflows using AI Agent Tool are executing the connected tools twice simultaneously.

What is the error message (if any)?

Setup:

  • Main agent: AI Agent v3 (orchestrator)

  • Sub-agent: AI Agent Tool v2.2 (called as a tool)

  • Memory: Redis Chat Memory connected to the sub-agent

Problem: When the orchestrator calls the sub-agent tool, the workflow tool executes 2 times at the same moment with identical parameters, creating duplicate records in my database with milliseconds of difference.

Please share your workflow

Main Workflow: Orchestrator (AI Agent v3)
  └── calls → Sub-agent (AI Agent Tool v2.2)
      └── executes → Workflow Tool (get_create_request)
          └── creates record in Supabase

Result: 2 identical records created instead of 1

Share the output returned by the last node

The workflow tool returns success twice, creating 2 identical database records with the same client_id but different request_id values, created milliseconds apart.

Information on your n8n setup

    1. n8n version: 2.4.8
    2. Database (default: SQLite): PostgreSQL (Supabase)
    3. n8n EXECUTIONS_PROCESS setting (default: own, main): main
    4. Running n8n via (Docker, npm, n8n cloud, desktop app): n8n cloud
    5. Operating system: Cloud
2

hey, yeah that’s a frustrating one. the duplicate execution after the n8n update is likely coming from how the AI Agent Tool handles tool calls when there’s a memory node connected.

here’s what i’d try:

**quick diagnostic first:**

1. disconnect the Redis Chat Memory from the sub-agent temporarily and see if the duplicates stop

2. if they do, the issue is the memory integration - if not, it’s something else with the agent setup

**if it’s the memory issue** (most likely):

the AI Agent Tool v2.2 sometimes has a race condition where it calls the tool, the memory node processes the result, and then something triggers a retry. this got worse in recent updates.

**quick fixes to try:**

1. make sure you’re on the latest AI Agent Tool version - there were fixes for this

2. in the sub-agent’s settings, check if there’s a “retry on error” or duplicate execution setting thats enabled. if so, disable it

3. try using AI Agent Tool v3 instead of v2.2 (i know its a subagent but the v3 is more stable with memory)

**if you still get duplicates:**

the redis memory might be causing issues with the agent’s internal state. try these:

- clear the redis cache completely

- check if the same chat session ID is being passed to both the orchestrator and sub-agent (sometimes this causes self-loops)

- verify the workflow tool itself isnt set to execute twice (check the node settings, not just the connection)

**docs to check:**

https://docs.n8n.io/integrations/builtin/cluster-nodes/n8n-nodes-langchain_memory_redis/

https://docs.n8n.io/integrations/builtin/cluster-nodes/n8n-nodes-langchain_agent/

lmk which of these you find - the memory disconnect test will tell us pretty quick if thats the culprit. duplicates on agent tools after updates usually comes down to memory handling or retry logic going weird.