Agents are not responding for prompts
istvanrma
In the last few days I notice almost none of my agents are working (that I created prior or recently).
They cannot even answer a simple question "Can you summarise what is in your instruction?" It used to answer it right away. Repeatedly deleted and recreated the agent, log out & in to Taskade, restarted browser but same issue occurs.
A week ago I were able to automate task complete / Ask Agent / Create Project now it is or time out or say "Ask Agent Failed: 400 This Model's Maximum Context Length Is 128000 Tokens. However, Your Messages Resulted In 769997 Tokens (769928 In The Messages, 69 In The Functions). Please Reduce The Length Of The Messages Or Functions."
but it is the same Agent that created for me many form of content in the past.
When it does work, for example the new feature Multi agent mode in the Project /and call agent do not remember any of the customisation (neither the custom instruction nor the knowledge files.
Log In
r
ryantaskade
Hi there,
Thank you for bringing this to our attention. Here are some possible explanations to the errors you have been facing:
- We utilize GPT 4.5 Turbo to power our agents, there was some downtime experienced 5 days ago which might explain the network request failed error your received
- The model's context window is capped at 128k tokens, the error message tells you that your request consumed 769,997 tokens, tokens is a measure of how large/complex your request to the AI is, this is computed by taking into account the prompt, any knowledge sources and context of the request. This means that the knowledge given to the agent is too large or the prompt given to the agent is too complex.
I'm unsure as to what you are referring to by "Home agent", could you please share more? Assuming you are comparing the agent and the default "AI Assistant" the request to the generic assistant doesn't consume as much tokens as the one with knowledge sources and context.
There isn't a definite way to limit the knowledge sources because a file of the same size (for example 2mb) can contain vastly different amount of "knowledge" for the agents and therefore have request that consume different number of tokens.
istvanrma
Hey ryantaskade,
Thanks to get back to me.
I wrote yesterday a long message while I was testing the Taskade system. Wanted to say until 'now' (yesterday) the issues still stands.
However, I have noticed the system update and since most of the agents started working again. I did not do automation or any hard task yet.
Narek Zograbian
istvanrma: Are you still experiencing this issue?
istvanrma
Forgot to say the Automation that used to work with another agent it doesn't work. Times out, stuck, give error code (I think error 400 or 500) . one time said " "Ask Agent Failed: 400 This Model's Maximum Context Length Is 128000 Tokens. However, Your Messages Resulted In 769997 Tokens (769928 In The Messages, 69 In The Functions). Please Reduce The Length Of The Messages Or Functions."
What can be the problem?
To much data point in Knowledge base?
Thogh some of my Agents, has nothing in its knowledge (neither added Project nor Media files) and still stuck or time out.
Thinking out loud, Somewhat the first edition Agents works better created at HOME. and under Projects and in Folders they fail?!
John at Taskade
istvanrma are you still experiencing issues? If so please let us know, any additional details would be helpful, thank you.
CC Narek Zograbian
peterkarlen
John at Taskade I'm also experiencing this issue in some projects.
Narek Zograbian
peterkarlen: Are you still experiencing it?