Coding with an AI assistant

Give it the spec

The OpenAPI spec at https://quova.io/openapi.json lists every endpoint with its parameters, response fields and credit cost, in a format AI assistants read well. Paste the link into your prompt, or download the file and attach it.

If your assistant can browse, you can also point it at these docs. The getting started guide explains the response format, errors and credits in one page.

Some assistants can also call Quova directly while you work, which helps when you want to check what the data looks like before writing code. See AI agents and MCP to set that up.

An example prompt

A prompt that names the spec, where the key lives and how to handle errors gives the assistant what it needs.

Write a Python script that uses the Quova API to get the dividends
paid by ASX:VAS and ASX:VGS over the last 12 months, then print the
total franking credits for each.

The OpenAPI spec is at https://quova.io/openapi.json
and the docs are at https://quova.io/docs/.

Read my API key from the QUOVA_API_KEY environment variable and send
it as a bearer token. Handle the errors array in the response, since
one symbol can fail while the others succeed.

What to check in the code

Check that the code reads the key from an environment variable or a secrets store. Keys shouldn't be pasted into the code, and they shouldn't go in a chat either.

Look for handling of the errors array. When you ask for several symbols, one can fail while the rest succeed, and code that only reads data will quietly skip it.

If the code loops over symbols one at a time, ask the assistant to send them in a single request. Most endpoints take up to 100 symbols per call, which is faster and uses the same credits.