Response exploration in Codex
The agent reads the real response before it writes the code that parses it.
The gap that costs the most time is between what you think a response contains and what it contains. Docs describe the general case, and the field you need is optional on the one account you tested with. Because the call and the code land in the same session, Codex can call the endpoint, look at the actual payload, and write the parsing against it — including the fields that come back null.
- 1A KonbiniAPI accountThe free plan includes 100 credits, which is enough to run this a few times.
- 2Codex connectedOAuth or API key.
- 3A real targetA public account, post, or keyword to run it against — the placeholders below are not live values.
Build with it.
Ask Codex for what you need. It calls the tool, sees the real response, and writes code against the shape that came back rather than one it guessed — each call costs the same credits as the matching REST call.
Good for
- See the real shape of an endpoint before writing anything against it
- Find out which fields are actually populated for the accounts you care about
- Build fixtures from responses that really happened
- Check what a normalized response looks like across two different platforms
Call the TikTok user endpoint for @[username] through MCP and show me the full response. Which fields are null?
Fetch the same profile from TikTok and Instagram and show me where the normalized shapes differ.
Get a Reddit post with its comments and map out the nesting so I know how deep replies go.
Call the LinkedIn company endpoint and write a TypeScript type that matches exactly what came back.
Fetch one page of results, then the next, and show me how the cursor works in practice.
Swap the bracketed parts for your own project — the platform, the endpoint, and where the result should land.
Run this in Codex.
One API key works for MCP and the REST API. The free plan includes 100 credits and no credit card.