Prompts library

The API Agent prompts library provides a comprehensive guide on using prompts to complete various tasks.

Guidelines for using prompts

To achieve accurate and efficient results, use the following guidelines when you enter the prompt:
  1. Avoid the words or tone that convey sarcasm or humor, as they can be misinterpreted.
  2. Use active voice, correct grammar, spelling, and keep your request clear and specific.
  3. Provide clear context and relevant keywords, and avoid vague or domain-specific terms. For example, generate test or build API.
  4. If the system is unable to understand your request, it offers suggestions for the requests.

Prerequisites

The API Agent plug-in can be used to complete tasks that are related to API Connect.

  • Ensure that you are logged in to API Agent. For login and operation instructions, see Getting started.
  • You must have valid permissions to use the API Agent. For more information, see API Agent user roles.
  • Ensure that you have the same permissions through the API Connect’s API Manager interface. Even if you have permission to use the API Agent plug-in, actions fail if you do not have the necessary permissions to complete the same tasks through the API Manager UI.

Usage of prompts for different cases

The following table explains the prompts and its usage for different actions.
Table 1. Prompts usage
Action Prompt Description
Generating test cases

Generate test cases for the API

Generate test cases for <pulbished api name>
  • Check that the API is published before you generate test cases for it. For more information, see Creating and publishing draft API.
  • API Agent connects to smart generation feature of test module, which creates test cases iteratively for the first 15 endpoints in the available API file. The test cases are stored in a test suite that is suffixed withAA_Test_Suite along with api name (AA_Test_Suite_<api_name>).
  • After generation is completed, API Agent displays the status message that shows the number of test cases that are created and the name of the test suite where they were created.
  • You can find the generated test cases in the Test module of the API Manager console.
Running test cases

Run test cases for the API

Run test cases for <pulbished api name>
Note: The <published api name> must contain the name of the API. Do not combine the version information. For example, petstore.
  • You can use to run test cases for a published API in Visual Studio Code, including initiating the test, viewing results, and optionally deleting a test case.

  • The Smart Generation feature of the API Agent test module generates the data for input parameters to run the test. The API Agent displays the status message and the URLs of test results in Visual Studio Code after runs are initiated successfully.

    Click the hyperlink to view the results of the individual test cases in the API Manager console.
Deleting test cases 
delete test case <test_name> in suite <suite_name>
Listing test suites 
List test suites
 You can use to list test suits. Also, you can delete a test suite by entering a prompt with the suite name.
Deleting test suite 
delete test suite <suite_name>
Listing test cases 
List test cases for test suite <test suite name>
  • You can use to list test cases corresponding to a test suite.
  • Ensure that the test suites are available. If not available, follow the steps that are mentioned in the Generating test cases row.
  • To identify the available test suites, see the Listing test suites row.
Creating an OpenAPI
  • To create an API specification for orders table 
    create openapi for orders table
    or 
    generate api spec for orders table
    Note: The database must not contain another table with the same name in any other schema, as this situation might cause confusion.
  • To create an API specification for orders table 
    create openapi for orders table with schema_name as purchase_order
    Or 
    create openapi for orders table in purchase_order schema
  • To create an API specification for orders table by using object name and without schema name 
    create openapi for orders table with object_name as purchase_order_orders
  • To create an API specification for orders table by using source ID
    create openapi for orders table with source_id as
    6c115670-a7e1-11ef-b576-fa5e88d5392d
  • To extract the OpenAPI specification for the table name and transforms to code generator-specific format 
    Extract the openapi spec for the given table name and transforms to Code generator specific format
 The following are the examples of prompts that you can use to create an Open API
Generating code
  • To generate code for a simple server type 
    Generate some Python FastAPI server code based on the following openapi @my_file.yaml
    Or 
    Please build FastAPI backend code using the OpenAPI file @api_spec.yaml
    Or 
    Can you develop a Python FastAPI server using the openapi specification provided in @service_definition.json
  • To generate code for a database create, retrieve, update, and delete server type 
    Generate some Python FastAPI server code based on the following openapi @my_file.yaml that handles database operations
    Or 
    Can you develop a database driven Python FastAPI server using the openapi specification provided in @service_definition.json
    Note: To know more information about usage of prompts by using OpenAPI Python Generator tool, see OpenAPI Python Generator tool.
  
[Technology preview] MCP server
  • To view the list of enabled servers 
    List me all the MCP servers that are currently enabled
    Or 
    List all available MCP servers
  • To get the details of the MCP server 
    Fetch the details of GitHub™ MCP server
  • To update the MCP server 
    Update the enabled MCP server gitlab™ with updated deployed url [Add URL]
  • To disable the MCP server from GitHub
    Remove the github™ MCP server from the enabled tools
  • To retrieve detailed information about a specific MCP server 
    Fetch details of the mcp server postman™
    or 
    Show me the configuration details for the github™ MCP server
    Or 
    Get information about the weather MCP server
  • To check whether an MCP server exists or not 
    Check the existence of mcp server Github™
    Or 
    check availability of mcp server postman™
    Or 
    check if postman™ mcp server is available
 Use the following prompts to do the following actions with the MCP server:
  • List of enabled servers
  • Server details
  • Update
  • Disable
  • Details about a specific MCP server
  • Check existence of MCP server

Sequential prompts

You can use API Agent to chain multiple tools together in a single prompt and then run one after another. The following are examples of sequential prompts:
  1. To create and run test cases 
    create test cases for the petstore api and run the test cases

  2. To list draft or published APIs or products 
    list draft apis, draft products, published apis and published products
    Or 
    list published apis

  3. To create a draft API and generate FastAPI server code 
    create draft api from file @openapi.yaml and generate fastapi server code from the same api file
    Or 
    create draft api using @file_name
  4. To review security of an API 
    review security of api API_NAME:VERSION
  5. To review security and validate an API 
    review security of api API_NAME:VERSION
    Or 
    validate api API_NAME:VERSION