Skip to main content
The secrets command helps you set up and manage API keys required for MCPKit to function.

​
Usage

mcpkit secrets [subcommand]

​
Subcommands

​
(none)
default
Run the interactive setup wizard to configure or update your API keys.
​
show
string
Display your current configuration (without revealing full API keys).

​
Examples

​
Initial Setup

Run the interactive setup wizard: 
mcpkit secrets
You’ll be prompted for:
  1. Browserbase API Key 
    ? Enter your Browserbase API key: bb_xxxxxxxxxxxxxxxx
  2. LLM Provider 
    ? Select your LLM provider:
❯ Google Gemini (recommended)
  OpenAI
  Anthropic
  Azure OpenAI
  3. LLM API Key 
    ? Enter your Gemini API key: AIxxxxxxxxxxxxxxxxx
Your secrets are now configured and saved to ~/.mcpkit/secrets.json

​
View Current Configuration

Check what’s currently configured: 
mcpkit secrets show
Output: 
Current Configuration:
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
LLM Provider: google
LLM Model: gemini-2.0-flash-exp
Browserbase API: Configured ✓
LLM API Key: Configured ✓

​
Configuration Details

​
Storage Location

Your secrets are stored in a JSON file at:
  • macOS/Linux
  • Windows
~/.mcpkit/secrets.json
Never commit secrets.json to version control. This file contains sensitive API keys.

​
File Format

The secrets file has this structure: 
{
  "browserbaseApiKey": "bb_xxxxxxxxxxxxxxxx",
  "llmProvider": "google",
  "llmApiKey": "AIxxxxxxxxxxxxxxxxx",
  "llmModel": "gemini-2.0-flash-exp"
}

​
Supported LLM Providers

MCPKit supports multiple LLM providers for AI-powered action discovery:

Google Gemini

Best balance of speed, quality, and cost. Free tier available.
Get an API key: Google AI Studio Supported models:
  • gemini-2.0-flash-exp (recommended)
  • gemini-1.5-pro
  • gemini-1.5-flash
Configuration: 
{
  "llmProvider": "google",
  "llmApiKey": "AIxxxxxxxxxxxxxxxxx",
  "llmModel": "gemini-2.0-flash-exp"
}

​
OpenAI

OpenAI

High quality but more expensive. Good for complex websites.
Get an API key: OpenAI Platform Supported models:
  • gpt-4o
  • gpt-4o-mini
  • gpt-4-turbo
Configuration: 
{
  "llmProvider": "openai",
  "llmApiKey": "sk-xxxxxxxxxxxxxxxx",
  "llmModel": "gpt-4o"
}

​
Anthropic

Anthropic

Excellent reasoning capabilities for complex page analysis.
Get an API key: Anthropic Console Supported models:
  • claude-3-5-sonnet-20241022
  • claude-3-opus-20240229
Configuration: 
{
  "llmProvider": "anthropic",
  "llmApiKey": "sk-ant-xxxxxxxxxxxxxxxx",
  "llmModel": "claude-3-5-sonnet-20241022"
}

​
Azure OpenAI

Azure OpenAI

Enterprise option with compliance and data residency.
Configuration: 
{
  "llmProvider": "azure",
  "azureApiKey": "xxxxxxxxxxxxxxxx",
  "azureEndpoint": "https://your-resource.openai.azure.com",
  "azureDeployment": "gpt-4o",
  "llmModel": "gpt-4o"
}

​
Browserbase Configuration

MCPKit uses Stagehand and Browserbase for serverless browser automation.

​
Get a Browserbase API Key

1

Sign up

Create a free account at browserbase.com
2

Navigate to settings

Go to your dashboard and click on Settings
3

Copy API key

Find your API key in the API Keys section
4

Configure mcpkit

Run mcpkit secrets and paste your API key when prompted

​
Browserbase Pricing

  • Free tier: 100 hours/month of browser time
  • Perfect for development and testing
  • Upgrade for production use

​
Updating Secrets

To update your configuration, simply run the setup wizard again: 
mcpkit secrets
Your existing values will be shown as defaults. Press Enter to keep them, or type new values to update.

​
Manual Configuration

You can also manually edit the secrets file:
  • macOS/Linux
  • Windows
# Edit with your preferred editor
nano ~/.mcpkit/secrets.json

# Or use vim
vim ~/.mcpkit/secrets.json
Make sure the JSON is valid after manual edits, or MCPKit may fail to read the configuration.

​
Environment Variables

You can also configure secrets via environment variables (useful for CI/CD): 
export BROWSERBASE_API_KEY="bb_xxxxxxxxxxxxxxxx"
export LLM_PROVIDER="google"
export LLM_API_KEY="AIxxxxxxxxxxxxxxxxx"
export LLM_MODEL="gemini-2.0-flash-exp"

mcpkit create https://example.com
Environment variables take precedence over the secrets file.

​
Troubleshooting

If you get “Invalid API key” errors:
  1. Verify the key - Make sure you copied it correctly
  2. Check provider - Ensure you’re using the right provider
  3. Test the key directly with the provider’s API
  4. Regenerate - Create a new API key if needed
If MCPKit can’t find your secrets:
  1. Run setup again: 
    mcpkit secrets
  2. Check file permissions: 
    ls -la ~/.mcpkit/
  3. Create directory manually: 
    mkdir -p ~/.mcpkit
mcpkit secrets
If you hit rate limits:
  1. Check your usage on the provider’s dashboard
  2. Upgrade your plan if needed
  3. Switch providers temporarily: 
    mcpkit secrets
# Select a different provider
If your secrets file is corrupted:
  1. Backup the file: 
    cp ~/.mcpkit/secrets.json ~/.mcpkit/secrets.json.bak
  2. Delete and recreate: 
    rm ~/.mcpkit/secrets.json
mcpkit secrets
  3. Validate JSON: 
    cat ~/.mcpkit/secrets.json | python -m json.tool

​
Security Best Practices

Protect Your API Keys

Your API keys provide access to paid services. Follow these practices to keep them secure:
  • ✅ Never commit secrets files to git
  • ✅ Use environment variables in CI/CD
  • ✅ Rotate keys regularly
  • ✅ Use separate keys for development and production
  • ✅ Set spending limits on provider dashboards
  • ❌ Don’t share secrets files
  • ❌ Don’t expose keys in logs or screenshots
  • ❌ Don’t use production keys in public repositories

​
Next Steps

Create MCP Server

Use your configured secrets to create an MCP server

Manage Contexts

Learn about authentication contexts