TOON format

TOON (Token Optimized Object Notation) is a compact response serialization for LLM consumption: same data as the JSON responses, at a fraction of the token count. If you are feeding API responses into a model's context window (agents, copilots, RAG over job lists), TOON stretches it substantially.

How to use it

Send Accept: application/toon on any endpoint:

bash

curl "https://api.aiveed.io/v1/jobs?limit=50" \
  -H "X-API-Key: $AIVEED_API_KEY" \
  -H "Accept: application/toon"

The response arrives as Content-Type: application/toon with an X-TOON-Dict: aiveed.1 header naming the abbreviation dictionary in use. Field names are shortened per the dictionary, repeated values are folded into enums, and list rows share one header — which is where the token savings come from.

When to use it

Use TOON when the consumer is a model: agents polling job lists, assistants summarizing usage.
Use JSON (the default) when the consumer is code: it is the stable, schema-documented contract.

Everything else — status codes, error codes, rate-limit headers — is identical in both formats.