Skip to main content
Patcha can use LLM providers to resolve complex vulnerabilities that can’t be fixed automatically.

​
Supported Providers

ProviderModelSetup Required
AnthropicClaudeAPI key
OpenAIGPT-4API key
GoogleGeminiAPI key

​
Setup

​
1. Install the package

Patcha is installed with LLM support by default: 
npm install -g patcha

​
2. Set the provider

patcha config set-provider anthropic

​
3. Configure API key

Using environment variable: 
patcha config set-api-key anthropic $ANTHROPIC_API_KEY
Or using direct value: 
patcha config set-api-key anthropic sk-ant-api03-xxxxx
Be careful when setting API keys directly. Use environment variables in CI/CD.

​
Anthropic (Claude)

patcha config set-provider anthropic
patcha config set-api-key anthropic $ANTHROPIC_API_KEY
Required environment variable: ANTHROPIC_API_KEY

​
OpenAI

patcha config set-provider openai
patcha config set-api-key openai $OPENAI_API_KEY
Required environment variable: OPENAI_API_KEY

​
Google Gemini

patcha config set-provider google
patcha config set-api-key google $GOOGLE_API_KEY
Required environment variable: GOOGLE_API_KEY

​
Using AI Fixes

Once configured, use the --ai flag: 
patcha fix --ai
This enables Level 3 (AI-assisted) resolution.

​
Disable AI Features

patcha config set-provider none
Or disable only Level 3: 
patcha config auto-fix --no-level3

​
Verify Configuration

patcha config list
Shows:
  • Current provider
  • Available providers
  • Auto-fix levels status

​
Cost Considerations

  • Level 1 and 2 are free (no API calls)
  • Level 3 uses API credits based on:
    • Number of vulnerabilities
    • Complexity of each case
    • Model used
Start with a dry run to estimate costs:
patcha fix --ai --dry-run