المجلدات

Missing or invalid API key.

{
  "error": {
    "code": "unauthorized",
    "message": "Invalid or missing API key."
  }
}

Two different refusals share this status, and a client must tell them apart by error.code. forbidden means the key's scope or allowlist does not permit this operation, or Pundit denied the action. plan_upgrade_required means the company's plan carries no API access at all, so no key on it can ever succeed and there is nothing to retry.

{
  "error": {
    "code": "forbidden",
    "message": "This API key is not permitted to perform this operation."
  }
}

{
  "error": {
    "code": "plan_upgrade_required",
    "message": "خطة «الاحترافي» لا تشمل الوصول إلى API. رقِّ إلى «المتطور» أو «الانتاجي» لتفعيله. — The الاحترافي plan does not include API access. Upgrade to «المتطور» or «الانتاجي» to enable it."
  }
}

Resource not found or not accessible — also returned for EVERY endpoint when the company's rest_api feature flag is disabled (the surface is hidden). Lookups drill through the URL hierarchy, so another company's record is a 404 and never a leak. A path that matches no route at all answers 404 with the distinct code unknown_endpoint and echoes the path back, so a mistyped or half-built URL is told apart from a record that is missing or out of reach.

{
  "error": {
    "code": "not_found",
    "message": "Resource not found, or you do not have access to it."
  }
}

A uniqueness/record conflict; retry.

{
  "error": {
    "code": "conflict",
    "message": "Could not complete due to a conflict; please retry."
  }
}

A caller-fixable bad request (validation error, bad date, bad enum).

{
  "error": {
    "code": "unprocessable_entity",
    "message": "Title can't be blank"
  }
}

Either the company's daily API allowance is spent (rate_limit_exceeded), or the per-key / per-IP burst throttle of 300 requests per minute fired. Both come back after a wait, so Retry-After is honest here. The two bodies are not the same shape. The daily-quota refusal uses the standard error envelope. The burst throttle is served by Rack::Attack ahead of the application, so its body is a flat {"error": "<string>"} with no code. A client that reads error.code has to tolerate error being a plain string.

{
  "error": {
    "code": "rate_limit_exceeded",
    "message": "استهلكت رصيد اليوم من طلبات API في خطة «المتطور» (1000 طلب يوميًا). يتجدد الرصيد عند منتصف الليل بتوقيت Asia/Riyadh. — Daily API quota exhausted: the المتطور plan allows 1000 calls per day. It resets at midnight Asia/Riyadh."
  }
}

{
  "error": "Rate limit exceeded. Please try again later."
}

Missing or invalid API key.

{
  "error": {
    "code": "unauthorized",
    "message": "Invalid or missing API key."
  }
}

Two different refusals share this status, and a client must tell them apart by error.code. forbidden means the key's scope or allowlist does not permit this operation, or Pundit denied the action. plan_upgrade_required means the company's plan carries no API access at all, so no key on it can ever succeed and there is nothing to retry.

{
  "error": {
    "code": "forbidden",
    "message": "This API key is not permitted to perform this operation."
  }
}

{
  "error": {
    "code": "plan_upgrade_required",
    "message": "خطة «الاحترافي» لا تشمل الوصول إلى API. رقِّ إلى «المتطور» أو «الانتاجي» لتفعيله. — The الاحترافي plan does not include API access. Upgrade to «المتطور» or «الانتاجي» to enable it."
  }
}

Resource not found or not accessible — also returned for EVERY endpoint when the company's rest_api feature flag is disabled (the surface is hidden). Lookups drill through the URL hierarchy, so another company's record is a 404 and never a leak. A path that matches no route at all answers 404 with the distinct code unknown_endpoint and echoes the path back, so a mistyped or half-built URL is told apart from a record that is missing or out of reach.

{
  "error": {
    "code": "not_found",
    "message": "Resource not found, or you do not have access to it."
  }
}

Either the company's daily API allowance is spent (rate_limit_exceeded), or the per-key / per-IP burst throttle of 300 requests per minute fired. Both come back after a wait, so Retry-After is honest here. The two bodies are not the same shape. The daily-quota refusal uses the standard error envelope. The burst throttle is served by Rack::Attack ahead of the application, so its body is a flat {"error": "<string>"} with no code. A client that reads error.code has to tolerate error being a plain string.

{
  "error": {
    "code": "rate_limit_exceeded",
    "message": "استهلكت رصيد اليوم من طلبات API في خطة «المتطور» (1000 طلب يوميًا). يتجدد الرصيد عند منتصف الليل بتوقيت Asia/Riyadh. — Daily API quota exhausted: the المتطور plan allows 1000 calls per day. It resets at midnight Asia/Riyadh."
  }
}

{
  "error": "Rate limit exceeded. Please try again later."
}