Errors

Every status code ACAAS can return, what it means, and how to recover.

ACAAS uses standard HTTP status codes. Successful requests return 200. Anything else falls into one of the categories below. All error responses are JSON.

Error response shape

Validation errors follow FastAPI’s standard envelope: a top-level detail array describing each problem.

{
  "detail": [
    {
      "loc": ["body", "intensity"],
      "msg": "Input should be less than or equal to 10",
      "type": "less_than_equal"
    }
  ]
}

• loc — path to the offending field, from the request root.
• msg — human-readable message.
• type — machine-readable error type, suitable for branching logic.
• input (optional) — the rejected value, when the server can echo it back.
• ctx (optional) — extra context, such as the boundary that was violated.

Non-validation errors (401, 413, 429, 5xx) return a simpler shape with a single detail string.

{
  "detail": "Invalid API key"
}

Status codes

200 OK

The request succeeded. Every capitalization endpoint returns the shout response envelope; /v1/emphasize returns the emphasis envelope.

401 Unauthorized

The X-API-Key header is missing, malformed, or revoked. The demo key acaas_banned_demo_key always returns 401 — it had its bullhorn revoked.

{ "detail": "Invalid API key" }

How to recover. Confirm the header is present and spelled exactly X-API-Key. Confirm the value is non-empty. In production, rotate the key from the dashboard.

413 Payload Too Large

The text field exceeds 10,000 characters. ACAAS will not amplify what it cannot lift.

{ "detail": "Text exceeds 10000 character limit" }

How to recover. Chunk the input client-side, call /v1/shout (or any capitalization endpoint) per chunk, and concatenate the results. Case conversion is per-character, so chunked output equals single-call output.

422 Unprocessable Entity

The body is valid JSON but failed schema validation. Common causes:

• text is missing or empty.
• intensity (on /v1/scream) is outside 1..10.
• emphasis_ratio (on /v1/emphasize) is outside (0, 1].

{
  "detail": [
    {
      "loc": ["body", "text"],
      "msg": "String should have at least 1 character",
      "type": "string_too_short"
    }
  ]
}

How to recover. Read loc to find the bad field, surface msg to the user (or log it), and resubmit with a valid value. Treat type as a stable key for client-side branching.

429 Too Many Requests

You have exhausted your quota for the current rate limit window. ACAAS does not serve another request until the window resets.

{ "detail": "Rate limit exceeded" }

How to recover. Call /v1/rate-limits to see how long until the window resets, then retry after that interval. For guidance on graceful backoff, see Rate limiting.

5xx Server Errors

Something went wrong on the ACAAS side. These are rare but possible during deployments or upstream model incidents (the latter affecting /v1/emphasize specifically).

How to recover. Retry with exponential backoff — start at one second, double on each attempt, cap at sixty seconds. Stop after five attempts and surface the failure. Check the status page if retries continue to fail.

Handling errors in code

A defensive pattern that handles all the above:

Python

import time
import requests

def shout(text: str, *, api_key: str, max_attempts: int = 5) -> str:
    backoff = 1.0
    for attempt in range(max_attempts):
        resp = requests.post(
            "https://api.acaas.example.com/v1/shout",
            headers={"X-API-Key": api_key},
            json={"text": text},
        )

        if resp.status_code == 200:
            return resp.json()["result"]

        if resp.status_code in (401, 413, 422):
            # Client errors are not retryable.
            raise ValueError(resp.json().get("detail"))

        if resp.status_code == 429:
            # Honor the rate limit window.
            status = requests.get(
                "https://api.acaas.example.com/v1/rate-limits",
                headers={"X-API-Key": api_key},
            ).json()
            time.sleep(status["resets_in_seconds"])
            continue

        # 5xx — exponential backoff.
        time.sleep(backoff)
        backoff = min(backoff * 2, 60)

    raise RuntimeError("ACAAS unavailable after retries")

Next steps

Rate limiting

Quota mechanics, the status ladder, and graceful backoff.

Quickstart

Make your first ACAAS request in under a minute.