GPTs Action Schema Action requires Approval

GPT builders Plugins / Actions builders
1

I want to integrate tavily.com into my GPT. I keep getting the following error which I do not have enough information to debug. I copied in the yamal fioe

debug] Response received
{
“function_name”: “searchData”,
“domain”: “api.tavily.com”,
“message”: “The requested action requires approval”,
“action_id”: “g-88b37b889b4908f284862be51c41648ff56fa945”
}
Action Schema
openapi: 3.0.3
info:
version: 1.0.0
title: Randolph Hill Search System
termsOfService: Terms of Use | SmartBear Software
contact:
email: [my email]
license:
name: Apache 2.0
url: Apache License, Version 2.0
servers:

  • url: https://api.tavily.com
    paths:
    /search:
    post:
    operationId: searchData
    description: API will search the internet for information requested.
    parameters:
    - name: api_key
    in: query
    description: Your unique API key.
    required: true
    schema:
    type: string
    example: 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
    - name: query
    in: query
    description: The search query string.
    required: true
    schema:
    type: string
    example: Joe Biden meeting with President Xi Jinping
    - name: max_results
    in: query
    description: The maximum number of search results to return.
    required: false
    schema:
    type: integer
    example: 5
    responses:
    ‘200’:
    description: Successful response
    content:
    application/json:
    schema:
    type: object
    properties:
    query:
    type: string
    follow_up_questions:
    type: array
    items:
    type: string
    answer:
    type: string
    images:
    type: array
    items:
    type: string
    results:
    type: array
    items:
    type: object
    properties:
    title:
    type: string
    url:
    type: string
    content:
    type: string
    score:
    type: number
    raw_content:
    type: string
    response_time:
    type: number
    components:
    schemas: {}
2 Likes
2

i have the same problem … did you find a solution ?
Could it be some policy violation ?

2 Likes
3

Same here. I don’t even see any connection in the logs on my server.

Edit: I’ve tried use regular chat to access the API, and it appear that it use only Bing for connection (maybe it is only regular chat functionality and GPTs use stand alone crowler). But due to fact that domain is not present in bing it failed to access it. Also it wrote to me that not being able to read robots.txt means that by default it won’t process any link from this domain.

2 Likes
4

I’ve changed the domain of the api from .name1.com.mx to .name2.mx , pointing to another ip. and i’m getting still the same mistake , somehow discarding a blacklist on the domain name.

5

It seems the problem originated from the OpenAPI declaration. I rewrote the declaration using the weather example as a basis, and the problem was solved.

6

Can you provide your JSON schema? I got a Tavily assistant (using OpenAI and Tavily APIs) to work, but not the custom GPT using a schema.

7

I’m not using tavily, nor a JSON schema. I’m using a custom API, I’ve modified an example of OpenAPI given by the GPT builder

openapi: 3.1.0
info:
title: API de EventosB2B
version: 1.0.0
servers:

  • url: https: / / PERSONAL API URL
    paths:
    /EventosB2B_estandar:
    get:
    summary: “Obtiene una lista de eventos con filtros opcionales”
    operationId: “GetEventList”
    parameters:
    - in: query
    name: where
    schema:
    type: string
    description: “##Notes for the GPT##”
    - in: query
    name: limit
    schema:
    type: string
    description: “Limita la cantidad de resultados”
    responses:
    ‘200’:
    description: “Lista de eventos filtrados”
    content:
    application/json:
    schema:
    type: array
    items:
    $ref: ‘#/components/schemas/Evento’
    components:
    schemas:
    Evento:
    type: object
    properties: {} # Aquí se deben definir las propiedades del esquema Evento