Creating API Specification Using CurieTech AI

After some ups and downs, I finally found the right way to talk with CurieTech!

I remember trying CurieTech as soon as it was released. At first, apart from making some DataWeave transformations, I wasn’t very satisfied with its performance. Looking back, now I realize that my frustration came from CurieTech not giving me exactly what I wanted, but I guess the problem was in me.

Since then, CurieTech has improved — and so have I! 😄 Over time, we found a common language, and now I’m quite pleased with how smoothly our communication flows.

How to ask C.T. to write you API specification that you need ?

Well, I found that next steps are the best way to start communication.

Define basic points of specification in API Spec Generator:

1. API Name & Version

2. Base URI / Root Path

3. Resources (Entities)

4. Data Model (Request & Response Bodies)

5. Operations Needed

6. Response Codes

7. Security / Authentication

8. Examples (Optional but Helpful)

After defining everything above, my example looks like this:

API Name & Version
* my-first-acb-api v1
Base URI / Root Path
* /api/*
Resources (Entities)
* accounts, contacts, opportunities 
* nested resources are needed (e.g., /accounts/{accountId}/contacts)
Data Model (Request & Response Bodies)
* Provide 5 fields of your choice for each entity with types (string, number, boolean, date, etc.)
* Example for Account: { id: string, name: string, email: string, … }
CRUD Operations Needed
* Create → POST /{resource}
* Read (single + collection) → GET /{resource} and GET /{resource}/{id}
* Update → PUT /{resource}/{id} 
* Delete → DELETE /{resource}/{id}
Response Codes
* Typical: 200 OK, 201 Created, 400 Bad Request, 404 Not Found, 500 Internal Server Error
Security / Authentication
* Example: Basic Auth
Examples (Optional but Helpful)
    * Add Example request/response JSONs for each method recording what you wrote in data model. 
    * Do not write examples in quotes!

Additionally, place any traits, types, securitySchemes, and other reusable components in a separate folder. 
We want the API spec written according to best practices.
In the master branch, only define endpoints with !include for body and response examples.

After 5–10 minutes, you’ll have the API specification ready! Just copy and paste it into Design Center, and make sure to create separate folders and files for traits, types, and security.

Make sure to double-check that everything meets your requirements! From my experience, there may be a one or a few small adjustments, but nothing more than that.

If everything looks good, you’re ready to publish your first AI-written API specification to Exchange! If not, just continue your conversation with C.T., and you’ll quickly figure out how to get what you need.

Next, we’ll continue this journey by importing the API specification into Anypoint Code Builder in Visual Studio Code! Yes — I’m keeping up with the trends and stepping into the future! :)