Skip to content

API_USAGE.md

Latest commit

 

History

History
155 lines (123 loc) · 4.21 KB

API_USAGE.md

File metadata and controls

155 lines (123 loc) · 4.21 KB

API Documentation — StealthHumanizer v2

Endpoints

POST /api/humanize

Humanize AI-generated text with configurable style, tone, and multi-pass processing.

Request

{
  "text": "string (required) — Input text (max 10,000 words)",
  "level": "light | medium | aggressive | ninja (required)",
  "style": "academic | professional | casual | creative | technical (required)",
  "tone": "academic-formal | academic-casual | journalistic | ... (required)",
  "customTone": "string (optional) — Custom tone description",
  "model": "gemini | groq | openai | claude | ... (required)",
  "apiKey": "string (required) — Provider API key",
  "targetScore": "number (optional, default 80) — Target human score 50-100",
  "language": "string (optional, default 'auto') — Language code"
}

Response

{
  "sentences": [
    {
      "original": "string",
      "humanized": "string",
      "alternatives": [],
      "index": 0,
      "detectionScore": 75
    }
  ],
  "fullText": "string — Complete humanized text",
  "model": "gemini",
  "modelName": "Google Gemini",
  "wordCount": { "input": 150, "output": 162 },
  "timestamp": 1711891200000,
  "passes": 2,
  "finalScore": 87,
  "options": { "level": "ninja", "style": "academic", "tone": "conversational" }
}

Errors

Status Error Cause
400 Missing required fields text, model, or apiKey not provided
400 Exceeds 10,000 word limit Input too long
401 API error Invalid or expired API key
429 Rate limited Provider rate limit exceeded
500 Internal error Provider outage or unexpected error

POST /api/alternative

Get alternative rewrites for a specific sentence.

Request

{
  "original": "string (required) — Original sentence",
  "current": "string (required) — Current humanized version",
  "level": "string (required)",
  "style": "string (required)",
  "tone": "string (required)",
  "model": "string (required)",
  "apiKey": "string (required)"
}

Response

{
  "alternatives": [
    "First alternative sentence...",
    "Second alternative sentence...",
    "Third alternative sentence..."
  ]
}

Provider Configuration

All providers use the same request format. Keys are passed from the client in each request (never stored server-side).

Provider Endpoints

Provider Endpoint Protocol
Gemini generativelanguage.googleapis.com/v1beta Google AI REST
OpenAI api.openai.com/v1 OpenAI API
Claude api.anthropic.com/v1 Anthropic API
Groq api.groq.com/openai/v1 OpenAI-compatible
Mistral api.mistral.ai/v1 OpenAI-compatible
Cohere api.cohere.ai/v1 Cohere API
Together api.together.xyz/v1 OpenAI-compatible
OpenRouter openrouter.ai/api/v1 OpenAI-compatible
Cerebras api.cerebras.ai/v1 OpenAI-compatible
DeepInfra api.deepinfra.com/v1/openai OpenAI-compatible
HuggingFace api-inference.huggingface.co HF Inference API
Cloudflare api.cloudflare.com/client/v4 Workers AI

Adding a Custom Provider

Any OpenAI-compatible API can be used by modifying the provider list in lib/providers.ts:

{
  id: 'my-provider',
  name: 'My Custom Provider',
  apiUrl: 'https://my-api.com/v1/chat/completions',
  defaultModel: 'my-model',
  models: ['my-model', 'my-model-v2'],
  // ... rest of config
}

Then add a case in the generateWithProvider function using the openAICompatibleGenerate helper.

Rate Limiting

Rate limiting is handled by each provider. Approximate free tier limits:

Provider Free Limit Notes
Gemini 15 req/min, 1500/day Most generous
Groq 30 req/min Ultra-fast
OpenRouter Varies by model Some models free
Together 60 req/min Free tier limited
Cerebras 30 req/min Fast inference
Mistral 1 req/min (free) Rate limited
Cohere 10 req/min Trial rate

Security Notes

  • API keys are passed from the client in each request
  • Keys are NOT stored on the server
  • Keys are NOT logged
  • All provider communication is over HTTPS
  • No user data is persisted server-side