Skip to main content

​
Overview

Bring Your Own Key (BYOK) allows you to use your personal API keys from AI providers instead of WhatDoc’s shared quota. This unlocks:
  • Unlimited generation: No tier limits
  • Advanced models: Access to GPT-4, Claude, and Gemini Pro
  • Cost control: Pay only for what you use
  • Privacy: Keys are never stored in the database

​
How BYOK Works

When you provide a custom API key during project configuration:
  1. The key is stored in your browser’s localStorage
  2. On project creation, the key is sent via HTTP headers:
    • x-custom-api-key: Your API key
    • x-target-model: Selected model ID (e.g., gemini-2.5-pro)
  3. The backend relays the key to the AI provider for generation
  4. The key is never persisted in the WhatDoc database
  5. After generation, the key is discarded from the server
Your API key stays in your browser and is sent directly to the AI provider. WhatDoc acts as a secure relay without storing sensitive credentials.

​
Supported Providers

​
Google Gemini

Free Models (no key required):
  • gemini-2.5-flash-lite (default, fastest)
  • gemini-2.5-flash
Premium Models (requires API key):
  • gemini-2.5-pro
Get Your Key:
  1. Visit Google AI Studio
  2. Sign in with your Google account
  3. Click Create API Key
  4. Copy the key (format: AIzaSy...)
Pricing: Gemini offers a generous free tier with rate limits. Check Google AI pricing.

​
OpenAI

Available Models:
  • gpt-4o
  • gpt-4o-mini
Get Your Key:
  1. Visit OpenAI Platform
  2. Sign in or create an account
  3. Click Create new secret key
  4. Copy the key (format: sk-proj-...)
Pricing: OpenAI charges per token. GPT-4o costs ~2.50−2.50-10 per million input tokens. Check OpenAI pricing.
OpenAI requires billing information to use API keys. Ensure you have a payment method on file.

​
Anthropic

Available Models:
  • claude-3.7-sonnet
  • claude-3.5-haiku
Get Your Key:
  1. Visit Anthropic Console
  2. Sign in or create an account
  3. Click Create Key
  4. Copy the key (format: sk-ant-...)
Pricing: Claude charges per token with tiered pricing. Check Anthropic pricing.

​
Using BYOK

1

Import a Repository

Start by importing a repository from /import as usual.
2

Configure Project

On the /configure page, scroll to the Model & API Key section.
3

Select a Model

Choose your desired model from the dropdown:
  • Gemini: Free models available without key
  • OpenAI: Requires API key for all models
  • Anthropic: Requires API key for all models
Premium models are marked with a lock icon if no key is provided.
4

Enter Your API Key

Paste your API key into the API Key field.The placeholder text changes based on the selected provider:
  • Gemini: AIzaSy... (leave blank for free models)
  • OpenAI: sk-proj-... (API Key required)
  • Anthropic: sk-ant-... (API Key required)
Your key is validated by length (>20 characters) but not authenticated until generation starts.
5

Deploy Documentation

Click Deploy Documentation. Your key is used for this generation.The system bypasses tier limits when a valid custom key is detected:
const isCustomKeyValid = rawCustomKey && rawCustomKey !== 'null' && rawCustomKey.length > 20;

​
Model Selector Component

The model selector dynamically adjusts based on whether you’ve provided a key: 
<ModelSelector
  hasCustomKey={customKey.trim().length > 20}
  selectedModel={customModel}
  setSelectedModel={setCustomModel}
  onRequestKey={(model) => {
    // Prompts user to provide a key for premium models
  }}
/>
Behavior:
  • If hasCustomKey is false, premium models show a lock icon and trigger onRequestKey on selection
  • If hasCustomKey is true, all models are unlocked

​
Security & Privacy

Your API key is NEVER stored in the WhatDoc database.
The BYOK flow ensures maximum security:
  1. localStorage: Key is stored locally in your browser
  2. HTTPS Headers: Key is transmitted securely via x-custom-api-key header
  3. Server Relay: WhatDoc relays the key to the AI provider’s API
  4. No Persistence: The key is discarded after the request completes
  5. No Logs: Keys are not logged or stored in any backend system

​
Backend Validation

On the server, keys are validated but never saved: 
const rawCustomKey = (req.headers['x-custom-api-key'] || '').trim();
const isCustomKeyValid = rawCustomKey && rawCustomKey !== 'null' && rawCustomKey.length > 20;

const byokOptions = {
  customKey: isCustomKeyValid ? rawCustomKey : '',
  targetModel: req.headers['x-target-model'] || 'gemini-2.5-flash-lite',
};

// Passed to worker, never stored in DB
await docGenerationQueue.add('generateDocs', {
  projectId,
  repoUrl,
  commitHash,
  llmProvider,
  byokOptions,
});

​
Cost Estimation

​
Gemini

  • Free Tier: 15 requests per minute, 1,500 requests per day
  • Paid Tier: 0.075per1Minputtokens(Flash),0.075 per 1M input tokens (Flash), 1.25 per 1M input tokens (Pro)
Typical documentation generation:
  • Small repo (1,000 files): ~500K tokens → 0.04(Flash)or0.04 (Flash) or 0.63 (Pro)
  • Large repo (10,000 files): ~5M tokens → 0.38(Flash)or0.38 (Flash) or 6.25 (Pro)

​
OpenAI

  • GPT-4o: 2.50per1Minputtokens,2.50 per 1M input tokens, 10 per 1M output tokens
  • GPT-4o-mini: 0.15per1Minputtokens,0.15 per 1M input tokens, 0.60 per 1M output tokens
Typical documentation generation:
  • Small repo: ~500K tokens → 1.25(4o)or1.25 (4o) or 0.08 (4o-mini)
  • Large repo: ~5M tokens → 12.50(4o)or12.50 (4o) or 0.75 (4o-mini)

​
Anthropic

  • Claude 3.7 Sonnet: 3per1Minputtokens,3 per 1M input tokens, 15 per 1M output tokens
  • Claude 3.5 Haiku: 0.25per1Minputtokens,0.25 per 1M input tokens, 1.25 per 1M output tokens
Typical documentation generation:
  • Small repo: ~500K tokens → 1.50(Sonnet)or1.50 (Sonnet) or 0.13 (Haiku)
  • Large repo: ~5M tokens → 15(Sonnet)or15 (Sonnet) or 1.25 (Haiku)
Actual costs vary based on repository size, code complexity, and output length. These are rough estimates.

​
Tier Limits with BYOK

When using a custom API key, WhatDoc bypasses tier limits:
  • Free Tier: No generation count increment
  • Pro Tier: No repository limit enforcement
  • Rate Limits: Subject only to your API provider’s limits
You can generate unlimited documentation as long as you have API quota with your provider.

​
Removing Your API Key

To stop using BYOK:
  1. Clear the API Key field on the /configure page
  2. Click Deploy Documentation
  3. Or clear it from localStorage: 
    localStorage.removeItem('wtd_custom_key');
Future generations will use WhatDoc’s shared quota and be subject to tier limits.

​
Troubleshooting

IssueCauseSolution
”API Key required”Selected premium model without keyEnter your API key or switch to free model
Generation fails with 401Invalid or expired API keyVerify key is correct and has billing enabled
Generation fails with 429Rate limit exceededWait and retry, or upgrade your API plan
Key not recognizedLess than 20 charactersEnsure you copied the full key
Model unavailableProvider outageCheck provider status page

​
Debugging BYOK

If generation fails with a custom key:
  1. Verify Key: Test the key directly via provider’s API playground
  2. Check Billing: Ensure your provider account has active billing
  3. Check Quota: Verify you haven’t exceeded rate limits
  4. Check Format: Ensure key matches expected format:
    • Gemini: AIzaSy...
    • OpenAI: sk-proj-...
    • Anthropic: sk-ant-...

​
Best Practices

Do:
  • Use free Gemini models for testing
  • Monitor your API usage via provider dashboards
  • Set up billing alerts to avoid surprise charges
  • Keep your API keys secure and never share them
  • Rotate keys periodically for security
Don’t:
  • Commit API keys to version control
  • Share keys in public forums or screenshots
  • Use keys on shared/public computers without clearing localStorage
  • Exceed your API budget (set spending limits)

​
What’s Next?

Generating Docs

Learn the full documentation generation process

Customizing Docs

Customize your documentation with logos and branding

Build docs developers (and LLMs) love

Get started for freeTalk to us