UnTXT Developer Portal

API Keys

Keys authenticate your API requests via the X-API-Key header. Contact raunak@sataklela.com to generate or revoke keys.

Usage (last 30 days)

Daily breakdown

Billing

Balances

Exemption code

Sataklela core devs only — redeem a Basic (waives billing) or AW (waives the Academic Weapon business-domain check) exemption code.

API Reference

Base URL: https://untxt.uk

Authentication

All API calls must include your API key in the X-API-Key header:

X-API-Key: untxt_your_key_here

POST /api/external/v1/detect

Run AI-text detection on a document. By default, returns the full Analysis verdict — the same Academic Weapon (M13) debate the webapp runs. Pass legacy: true for the cheaper, faster heuristic score other AI detectors show: fast flags, not adjudication.

Auth: X-API-Key  |  Billing: Analysis (default) is metered — see GET /api/external/v1/portal's apiAnalysisEnabled. legacy: true is 1 CVT per 75 words (min 1).

Breaking change: the old useM13 flag is removed. It used to blend a heuristic score with M13 (30/70) when set to true; that blend is gone. Analysis is now the unflagged default, and the old heuristic-only behaviour is legacy: true.

Request

{
  "text":    "string   // 50–100 000 characters, required",
  "legacy":  false,     // optional, default false — true runs the cheap heuristic score instead of Analysis
  "rewrite": null       // optional — a style name from /tharoorifier.json; admin-cleared access only
}

Response 200 — default (Analysis / M13)

{
  "requestId":  "uuid",
  "score":      0.0–1.0,      // M13 consensus AI likelihood
  "isAI":       boolean,       // true if score ≥ 0.20
  "confidence": "low|medium|high",
  "cvtCharged": number,        // 0 during free trial
  "isTrial":    boolean,
  "modules": {
    "m13": { "score": 0.0–1.0, "segments": number, "debateRounds": number, "apiError": bool }
  }
}

Response 200 — legacy: true

{
  "requestId":  "uuid",
  "score":      0.0–1.0,      // heuristic composite AI likelihood
  "isAI":       boolean,       // true if score ≥ 0.20
  "confidence": "low|medium|high",
  "cvtCharged": number,
  "isTrial":    boolean,
  "modules": {
    "syntacticRegularity": { "score": 0.0–1.0, "apiError": bool },
    "forensicStylometry":  { "score": 0.0–1.0, "apiError": bool },
    "advancedSyntactic":   { "score": 0.0–1.0, "apiError": bool, "requiresApiKey": bool }
  }
}

Response 200 — with rewrite set

Adds a rewritten field alongside whichever response above applies — admin-cleared accounts only. rewritten.demo is true when no Anthropic key is configured server-side and the text is canned placeholder output, not a real rewrite (not billed in that case).

{
  "rewritten": { "style": "Lexical Flourish", "text": "The proposition collapses…" }
}

Example

curl -X POST https://untxt.uk/api/external/v1/detect \
  -H "Content-Type: application/json" \
  -H "X-API-Key: untxt_your_key_here" \
  -d '{"text": "In conclusion, it is important to note that..."}'

# cheaper, faster heuristic score instead of Analysis
curl -X POST https://untxt.uk/api/external/v1/detect \
  -H "Content-Type: application/json" \
  -H "X-API-Key: untxt_your_key_here" \
  -d '{"text": "In conclusion, it is important to note that...", "legacy": true}'

GET /api/external/v1/balance

Check your current CVT balance and billing status.

Response 200

{
  "balance":       number,       // available CVT
  "isTrialActive": boolean,
  "trialEndsAt":   "ISO 8601 date | null"
}

Error codes

CVT billing

1 CVT ≈ 75 words. Purchase CVT through your UnTXT account — contact your account manager for bulk pricing.