Skip to content
Notifications
Clear all

Persistent 'Tool not found' error even after correct installation. Help?

2 Posts
2 Users
0 Reactions
4 Views
(@kellyd)
Trusted Member
Joined: 1 week ago
Posts: 40
Topic starter   [#8825]

Hey everyone, super excited to be here and diving into SuperAGI! I've been trying to set up a workflow to automate some of our project status updates, and I've hit a wall that I can't seem to get past, no matter what I try.

I followed the installation guide to the letter, and I'm trying to use the `ReadFile` tool as a simple test. The tool is definitely listed in my `config.yaml` under the tools section, and I can see it in the UI when I'm creating an agent. But every single time I run the agent, I get this super frustrating "Tool not found" error in the logs. It's like the system knows about the tool when I'm setting things up, but then forgets it exists when it's time to actually run.

I've tried restarting the entire stack, double-checking the YAML formatting (even used a validator), and making sure there are no typos. I'm coming from a background of using more visual no-code automators like Make or Zapier, so maybe I'm missing something fundamental about how SuperAGI loads these tools? Is there a specific directory or a cache I need to be aware of?

Has anyone else run into this? How does SuperAGI's tool loading compare to something like n8n where you just install an integration node? I'm really keen to get past this and start building some useful automations for my team, but this first hurdle is a real blocker. Any guidance would be massively appreciated!



   
Quote
(@laurap)
Trusted Member
Joined: 1 week ago
Posts: 42
 

That sounds incredibly frustrating, especially when you've checked the basics. The UI showing the tool and then the runtime not finding it points to a potential loading order or path issue, which is different from the plugin systems in tools like n8n.

It's good you checked the config YAML. Next, I'd look at the agent's specific log file around startup, not just the runtime logs, to see if there's a failed import for that tool. Sometimes a missing Python dependency for that specific tool can cause it to fail silently during the agent's initialization, even if it's listed.

Could you share the exact error line from the log? That often has the specific path or module name the system is trying and failing to load, which is the quickest clue.


Be kind, stay curious.


   
ReplyQuote