Stealthwriter API Documentation

Introduction

The Stealth Writer API lets you humanize AI-generated text and detect AI-written content from your own applications. Every endpoint accepts a JSON request body and returns a JSON response. All requests are made over HTTPS.

Authentication

Authenticate every request with your API token, sent as a Bearer token in the Authorization header. The token is tied to your subscription and looks like sk_live_…. Requests with a missing, invalid, or disabled token — or a token that is not attached to an active API subscription — are rejected with 401 Unauthorized.

Authorization: Bearer sk_live_your_api_token
Content-Type: application/json

Plan limits & quotas

Your API plan can define several usage limits, each checked on every request. When a limit is reached you receive a 429 Too Many Requests with code: "limit_exceeded"; the reason field tells you exactly which limit was hit so you can handle it programmatically.

Daily counters reset at midnight and the monthly word total resets on the first of each month (server time). All counters only include successful (200 OK) requests — calls that fail validation, hit a limit, or error out do not consume your quota.

Rewrites AI-generated text so it reads as natural, human-written content. Returns one or more humanized variations of the supplied text.

Limits: subject to words_per_input, humanizations_per_day and words_per_month. Exceeding any configured limit returns 429 limit_exceeded; limits your plan doesn't configure are not enforced. See Plan limits and Errors.

Body parameters

Request

curl -X POST https://stealthwriter.io/api/v1/humanize \
  -H "Authorization: Bearer sk_live_your_api_token" \
  -H "Content-Type: application/json" \
  -d '{
    "text": "The mitochondria is the powerhouse of the cell.",
    "n": 1,
    "level": 2,
    "style": "casual"
  }'

const res = await fetch("https://stealthwriter.io/api/v1/humanize", {
  method: "POST",
  headers: {
    "Authorization": "Bearer sk_live_your_api_token",
    "Content-Type": "application/json",
  },
  body: JSON.stringify({
    text: "The mitochondria is the powerhouse of the cell.",
    n: 1,
    level: 2,
    style: "casual",
  }),
});

const data = await res.json();

$client = new \GuzzleHttp\Client();

$response = $client->post('https://stealthwriter.io/api/v1/humanize', [
    'headers' => [
        'Authorization' => 'Bearer sk_live_your_api_token',
        'Content-Type'  => 'application/json',
    ],
    'json' => [
        'text'  => 'The mitochondria is the powerhouse of the cell.',
        'n'     => 1,
        'level' => 2,
        'style' => 'casual',
    ],
]);

$data = json_decode((string) $response->getBody(), true);

Response 200 OK

{
  "backend": "local",
  "sentences": [
    {
      "index": 0,
      "input_sentence": "The mitochondria is the powerhouse of the cell.",
      "outputs": [
        {
          "text": "Honestly, the mitochondria is what keeps the cell powered up.",
          "readability": 45.29,
          "label": 0
        }
      ]
    }
  ]
}

Response fields are passed through from the Stealth Writer engine and may include additional metadata.

Analyzes a passage as a whole and returns the likelihood that it was written by AI.

Limits: subject to words_per_input, words_per_month and the shared ai_scans_per_day quota (counts toward the same daily total as /detector/sentence). Exceeding any configured limit returns 429 limit_exceeded; limits your plan doesn't configure are not enforced. See Plan limits and Errors.

Body parameters

Request

curl -X POST https://stealthwriter.io/api/v1/detector/passage \
  -H "Authorization: Bearer sk_live_your_api_token" \
  -H "Content-Type: application/json" \
  -d '{
    "text": "Paste the passage you want to analyze here.",
    "detector": "normal"
  }'

const res = await fetch("https://stealthwriter.io/api/v1/detector/passage", {
  method: "POST",
  headers: {
    "Authorization": "Bearer sk_live_your_api_token",
    "Content-Type": "application/json",
  },
  body: JSON.stringify({
    text: "Paste the passage you want to analyze here.",
    detector: "normal",
  }),
});

const data = await res.json();

$client = new \GuzzleHttp\Client();

$response = $client->post('https://stealthwriter.io/api/v1/detector/passage', [
    'headers' => [
        'Authorization' => 'Bearer sk_live_your_api_token',
        'Content-Type'  => 'application/json',
    ],
    'json' => [
        'text'     => 'Paste the passage you want to analyze here.',
        'detector' => 'normal',
    ],
]);

$data = json_decode((string) $response->getBody(), true);

Response 200 OK

{
  "ai_prob": 0.9881510734558105,
  "human_prob": 0.0118489945307374,
  "label": "AI"
}

ai_prob and human_prob are floats between 0 and 1 (not percentages) and sum to 1. label is the engine's verdict for the passage as a whole.

Breaks the text into sentences and returns a per-sentence AI/human probability and label — useful for highlighting exactly which parts read as AI-generated.

Limits: subject to words_per_input, words_per_month and the shared ai_scans_per_day quota (counts toward the same daily total as /detector/passage). Exceeding any configured limit returns 429 limit_exceeded; limits your plan doesn't configure are not enforced. See Plan limits and Errors.

Body parameters

Request

curl -X POST https://stealthwriter.io/api/v1/detector/sentence \
  -H "Authorization: Bearer sk_live_your_api_token" \
  -H "Content-Type: application/json" \
  -d '{
    "text": "First sentence. Second sentence to check.",
    "detector": "normal"
  }'

const res = await fetch("https://stealthwriter.io/api/v1/detector/sentence", {
  method: "POST",
  headers: {
    "Authorization": "Bearer sk_live_your_api_token",
    "Content-Type": "application/json",
  },
  body: JSON.stringify({
    text: "First sentence. Second sentence to check.",
    detector: "normal",
  }),
});

const data = await res.json();

$client = new \GuzzleHttp\Client();

$response = $client->post('https://stealthwriter.io/api/v1/detector/sentence', [
    'headers' => [
        'Authorization' => 'Bearer sk_live_your_api_token',
        'Content-Type'  => 'application/json',
    ],
    'json' => [
        'text'     => 'First sentence. Second sentence to check.',
        'detector' => 'normal',
    ],
]);

$data = json_decode((string) $response->getBody(), true);

Response 200 OK

[
  { "label": "Human" },
  { "label": "AI" }
]

The response is a bare JSON array of per-sentence verdicts, in the order the sentences appear in text. Each item carries only a label; the sentence text itself is not echoed back.

Sentences shorter than 5 words are skipped — the engine will not score them. The array is therefore often shorter than your sentence count, and is [] when every sentence is too short (a 200 OK with an empty array is a valid result, not an error). Do not assume item n corresponds to your n-th sentence.

The detector value (easy / normal / strict) controls how aggressively sentences are flagged.

Errors

Errors come back with the matching HTTP status code in one of two JSON shapes. Most errors use a consistent envelope with success, code, reason and message. Body-validation errors (422) use Laravel's standard message + errors shape instead.

Standard error envelope

{
  "success": false,
  "code": "limit_exceeded",
  "reason": "words_per_input",
  "message": "This request has 5200 words, which exceeds your plan limit of 5000 words per request."
}

Examples

401 Unauthorized — missing, invalid, or disabled token:

{
  "success": false,
  "code": "unauthorized",
  "reason": "invalid-api-token",
  "message": "The API token is invalid or has been disabled."
}

429 Too Many Requests — the request would exceed your monthly word allowance:

{
  "success": false,
  "code": "limit_exceeded",
  "reason": "words_per_month",
  "message": "This request's 800 words would exceed your monthly limit of 50000 words (you have used 49500 this month)."
}

429 Too Many Requests — a configured plan limit was reached:

{
  "success": false,
  "code": "limit_exceeded",
  "reason": "ai_scans_per_day",
  "message": "You have reached your daily limit of 500 AI scans. Try again tomorrow."
}

422 Unprocessable — body validation failed (note the different shape):

{
  "message": "The text parameter is required.",
  "errors": {
    "text": ["The text parameter is required."],
    "n": ["The n parameter is required."]
  }
}

Status reference