Actions ResponseTooLargeError - use function result as input for next function

GPT builders Plugins / Actions builders
,
1

I created a C# .NET 8 WebApplication that provides an EmailController and uses Swagger.

I want to give MyGPT the ability to use those actions. So I took the schema from the swagger.yaml and configured an Action from it.

So now my GPT has the following API endpoints available:

image
image743×376 6 KB

In my Instructions I make clear that GPT can make use of multiple endpoints and take results from a previous api call as input for the next one.

Purpose of GPT:
To assist users in integrating various C# API endpoints effectively.

Knowledge field/area of expertise:
C# Programming, API Integration

Approach this task step-by-step, take your time and do not skip steps:
Task to be done:
.1. Understand user requests in detail to extract essential information.
.2. Plan the use of multiple C# API endpoints to create an effective solution for the request.
.3. Utilize outputs from one API call as inputs for others to streamline the workflow.
.4. Monitor and manage API data to avoid ResponseTooLargeError.
.5. Present the plan to the user in a clear, numbered list and await their confirmation before proceeding.
.6.Output format: Deliver results efficiently and only show final outcomes.

My goal is to be able to anwer request like
Give me the subjects of my 3 newest E-Mails
expecting a plan like:

  1. Get Mails
  2. Order Mails Descending by Date
  3. Get first 3 Mails from the ordered list

So 2 needs the result from 1 as input and 3 needs result from 2 as input

But unfortunately my GPT just tries to fetch all the mails (Step 1) and seems to create an output what is currently too much output causing it to run into Error ResponseTooLargeError.

>[debug] Calling HTTP endpoint
{
  "domain": "gg123214-12345.euw.devtunnels.ms",
  "method": "get",
  "path": "/Emails",
  "operation": "Emails_GET",
  "operation_hash": "b078324jj34h23k4gjh23423jh4234234",
  "is_consequential": false,
  "params": {}
}

>[debug] Response received
{
  "response_data": "ResponseTooLargeError"
}

Isn’t there a way to keep the data just inside the WebApplication and GPT only prints out the endresult?

Because the endresult should not be bigger than 50 Tokens in this case…

So please help me what am i doing wrong here?