AI Agents

Point an agent at one of these and it has the whole API: the plain-text llms.txt files for context windows, and the OpenAPI spec for typed client generation. Both are generated from the same source as these docs, so they never drift.

llms.txt

We publish llms.txt files — a standard for making a site readable by large language models — in two depths:

/llms.txt — a structured overview: the create call, transforms, plans, and where to go next. Small enough to keep in context.
/llms-full.txt — the complete API reference in one plain-text file: every endpoint, the full transform parameter list, the App API, and error handling.

Feeding an agent? Hand it /llms-full.txt — complete API coverage in a single context load, no crawling.

bash

# Fetch the full reference
curl https://img.pro/llms-full.txt

OpenAPI spec

The full OpenAPI 3.1 specification, as YAML:

/openapi.yaml — every endpoint with request and response schemas, the authentication schemes, error shapes with their action objects, and the URL transform surface as a documented operation.

Feed the spec URL to an agent, import it into an API client (Postman, Insomnia, Bruno), or run it through any OpenAPI generator to get a typed client and the full tool set (create, get, list, update, delete) straight from the spec.

bash

curl https://img.pro/openapi.yaml

Transforms (resize, convert, background removal) aren’t extra API calls — every response’s url is transformable in place by appending query params (e.g. ?size=m, ?format=webp, ?segment=foreground). The OpenAPI spec documents them as a single image-URL operation; the full list is in Transformations.

Building an app where your own users authenticate and own their images, rather than acting with a single key? See Building an app — hosted login, code exchange, and acting on your users’ behalf with one machine secret. It’s included in both /llms-full.txt and the OpenAPI spec.