Skip to content
kit

Configuration

Configuration

Configure kit’s AI PR reviewer for your team’s needs with flexible model selection, API key management, and configuration options.

Model Override via CLI

Override the model for any specific review without modifying your configuration:

Terminal window
kit review --model gpt-4.1-nano https://github.com/owner/repo/pull/123
kit review --model gpt-4.1 https://github.com/owner/repo/pull/123


# Short flag also works
kit review -m claude-sonnet-4-20250514 https://github.com/owner/repo/pull/123

Available Models

Free Local AI (Ollama)

Perfect for unlimited reviews without external API costs:

Terminal window
# Popular coding models
qwen2.5-coder:latest      # Excellent for code analysis
deepseek-r1:latest        # Strong reasoning capabilities
gemma3:latest             # Good general purpose
devstral:latest           # Mistral's coding model
llama3.2:latest           # Meta's latest model
codellama:latest          # Code-specialized Llama

Setup:

Terminal window
# 1. Install Ollama
curl -fsSL https://ollama.ai/install.sh | sh


# 2. Pull a model
ollama pull qwen2.5-coder:latest


# 3. Use with kit
kit review --model qwen2.5-coder:latest <pr-url>

OpenAI Models

Terminal window
# Budget options
gpt-4.1-nano             # Ultra-budget: ~$0.0015-0.004
gpt-4.1-mini             # Budget-friendly: ~$0.005-0.015
gpt-4o-mini              # Newer mini model


# Standard options
gpt-4.1                  # Good balance: ~$0.02-0.10
gpt-4o                   # Latest GPT-4 model
gpt-4-turbo              # Fast GPT-4 variant

Anthropic Claude

Terminal window
# Budget option
claude-3-5-haiku-20241022    # Fast and economical


# Recommended
claude-3-5-sonnet-20241022   # Excellent balance
claude-sonnet-4-20250514     # Latest Sonnet (recommended)


# Premium
claude-opus-4-20250514       # Highest quality

Google Gemini

Terminal window
# Ultra-budget
gemini-1.5-flash-8b         # ~$0.003 per review


# Standard options
gemini-2.5-flash            # Excellent value: ~$0.007
gemini-1.5-flash            # Fast and efficient
gemini-1.5-pro              # More capable
gemini-2.5-pro              # Latest pro model

API Key Setup

GitHub Token

Get from GitHub Settings → Developer settings → Personal access tokens

Required permissions:

  • repo (for private repositories)
  • public_repo (for public repositories)
  • pull_requests:write (to post comments)
Terminal window
export KIT_GITHUB_TOKEN="ghp_your_token_here"

LLM Provider API Keys

Anthropic Claude (Recommended):

Terminal window
export KIT_ANTHROPIC_TOKEN="sk-ant-your_key"

Get from: Anthropic Console

OpenAI GPT Models:

Terminal window
export KIT_OPENAI_TOKEN="sk-your_openai_key"

Get from: OpenAI Platform

Google Gemini:

Terminal window
export KIT_GOOGLE_API_KEY="AIzaSy-your_google_key"

Get from: Google AI Studio

Ollama (Local - No API Key Required):

Terminal window
# Just ensure Ollama is running
ollama serve

Configuration Files

Basic Configuration

Edit ~/.kit/review-config.yaml:

github:
  token: ghp_your_token_here
  base_url: https://api.github.com


llm:
  provider: anthropic  # or "openai", "google", "ollama"
  model: claude-sonnet-4-20250514
  api_key: sk-ant-your_key_here
  max_tokens: 4000
  temperature: 0.1


review:
  post_as_comment: true
  clone_for_analysis: true
  cache_repos: true
  max_files: 50

Provider-Specific Configurations

Anthropic Claude:

github:
  token: ghp_your_token_here
  base_url: https://api.github.com


llm:
  provider: anthropic
  model: claude-sonnet-4-20250514
  api_key: sk-ant-your_key_here
  max_tokens: 4000
  temperature: 0.1


review:
  post_as_comment: true
  clone_for_analysis: true
  cache_repos: true
  max_files: 50

OpenAI GPT:

github:
  token: ghp_your_token_here
  base_url: https://api.github.com


llm:
  provider: openai
  model: gpt-4.1
  api_key: sk-your_openai_key
  max_tokens: 4000
  temperature: 0.1


review:
  post_as_comment: true
  clone_for_analysis: true
  cache_repos: true
  max_files: 50

Google Gemini:

github:
  token: ghp_your_token_here
  base_url: https://api.github.com


llm:
  provider: google
  model: gemini-2.5-flash  # or gemini-1.5-flash-8b for ultra-budget
  api_key: AIzaSy-your_google_key
  max_tokens: 4000
  temperature: 0.1


review:
  post_as_comment: true
  clone_for_analysis: true
  cache_repos: true
  max_files: 50

Free Local AI (Ollama):

github:
  token: ghp_your_token_here  # Still need GitHub API access
  base_url: https://api.github.com


llm:
  provider: ollama
  model: qwen2.5-coder:latest  # or deepseek-r1:latest
  api_base_url: http://localhost:11434
  api_key: ollama  # Placeholder (Ollama doesn't use API keys)
  max_tokens: 4000
  temperature: 0.1


review:
  post_as_comment: true
  clone_for_analysis: true
  cache_repos: true
  max_files: 50

Priority Filtering Configuration

Default Priority Settings

github:
  token: ghp_your_token_here
  base_url: https://api.github.com


llm:
  provider: anthropic
  model: claude-sonnet-4-20250514
  api_key: sk-ant-your_key_here
  max_tokens: 4000
  temperature: 0.1


review:
  post_as_comment: true
  clone_for_analysis: true
  cache_repos: true
  max_files: 50
  # Optional: Set default priority filter
  priority_filter: ["high", "medium"]  # Only show important issues by default

Priority Configuration Examples

Security-focused configuration:

review:
  priority_filter: ["high"]  # Critical issues only
  post_as_comment: true
  clone_for_analysis: true
  cache_repos: true

General development workflow:

review:
  priority_filter: ["high", "medium"]  # Skip style suggestions
  post_as_comment: true
  clone_for_analysis: true
  cache_repos: true

Code quality/style reviews:

review:
  priority_filter: ["low"]  # Focus on improvements
  post_as_comment: true
  clone_for_analysis: true
  cache_repos: true

Default (show all priorities):

review:
  priority_filter: ["high", "medium", "low"]  # Same as omitting priority_filter
  post_as_comment: true
  clone_for_analysis: true
  cache_repos: true

Advanced Configuration Options

Repository Analysis Settings

review:
  post_as_comment: true
  clone_for_analysis: true
  cache_repos: true
  max_files: 50


  # Advanced settings
  max_file_size: 1048576      # 1MB max file size
  exclude_patterns:           # Files to ignore
    - "*.lock"
    - "package-lock.json"
    - "yarn.lock"
    - "*.min.js"
    - "dist/"
    - "build/"


  include_patterns:           # Only analyze these files (if specified)
    - "*.py"
    - "*.js"
    - "*.ts"
    - "*.java"
    - "*.go"


  analysis_timeout: 300       # 5 minute timeout
  retry_attempts: 3           # Retry failed requests

Multi-Provider Configuration

github:
  token: ghp_your_token_here
  base_url: https://api.github.com


# Default provider
llm:
  provider: anthropic
  model: claude-sonnet-4-20250514
  api_key: sk-ant-your_key_here
  max_tokens: 4000
  temperature: 0.1


# Alternative providers (for CLI override)
providers:
  openai:
    api_key: sk-your_openai_key
    api_base_url: https://api.openai.com/v1


  google:
    api_key: AIzaSy-your_google_key
    api_base_url: https://generativelanguage.googleapis.com/v1beta


  ollama:
    api_base_url: http://localhost:11434
    api_key: ollama


review:
  post_as_comment: true
  clone_for_analysis: true
  cache_repos: true
  max_files: 50

Custom Profile Defaults

github:
  token: ghp_your_token_here
  base_url: https://api.github.com


llm:
  provider: anthropic
  model: claude-sonnet-4-20250514
  api_key: sk-ant-your_key_here
  max_tokens: 4000
  temperature: 0.1


review:
  post_as_comment: true
  clone_for_analysis: true
  cache_repos: true
  max_files: 50


  # Default profile for all reviews
  default_profile: "company-standards"


  # Repository-specific profiles
  profile_overrides:
    "cased/frontend-app": "frontend-react"
    "cased/api-service": "backend-python"
    "cased/security-lib": "security-focused"

Environment Variable Overrides

You can override any configuration setting using environment variables:

Terminal window
# GitHub settings
export KIT_GITHUB_TOKEN="ghp_your_token"
export KIT_GITHUB_BASE_URL="https://api.github.com"


# LLM settings
export KIT_LLM_PROVIDER="anthropic"
export KIT_LLM_MODEL="claude-sonnet-4-20250514"
export KIT_ANTHROPIC_TOKEN="sk-ant-your_key"
export KIT_LLM_MAX_TOKENS="4000"
export KIT_LLM_TEMPERATURE="0.1"


# Review settings
export KIT_REVIEW_POST_AS_COMMENT="true"
export KIT_REVIEW_CACHE_REPOS="true"
export KIT_REVIEW_MAX_FILES="50"
export KIT_REVIEW_PRIORITY_FILTER="high,medium"

Configuration Validation

Test your configuration:

Terminal window
# Initialize configuration with guided setup
kit review --init-config


# Validate current configuration
kit review --validate-config


# Test with dry run
kit review --dry-run --model claude-sonnet-4 https://github.com/owner/repo/pull/123

Multiple Configuration Profiles

Team-Specific Configs

Terminal window
# Create team-specific config directories
mkdir -p ~/.kit/profiles/frontend-team
mkdir -p ~/.kit/profiles/backend-team
mkdir -p ~/.kit/profiles/security-team


# Frontend team config
cat > ~/.kit/profiles/frontend-team/review-config.yaml << EOF
llm:
  provider: openai
  model: gpt-4.1-mini
  api_key: sk-frontend-team-key
review:
  default_profile: "frontend-react"
  priority_filter: ["high", "medium"]
EOF


# Use specific config
KIT_CONFIG_DIR=~/.kit/profiles/frontend-team kit review <pr-url>

Project-Specific Configs

Terminal window
# In your project directory
mkdir .kit
cat > .kit/review-config.yaml << EOF
llm:
  provider: ollama
  model: qwen2.5-coder:latest
review:
  default_profile: "project-standards"
  max_files: 30
EOF


# Kit automatically uses project-local config if available
kit review <pr-url>

Cost Management Configuration

Budget Controls

llm:
  provider: anthropic
  model: claude-sonnet-4-20250514
  api_key: sk-ant-your_key_here
  max_tokens: 4000
  temperature: 0.1


  # Cost controls
  cost_limit_per_review: 0.50    # Maximum $0.50 per review
  monthly_cost_limit: 100.00     # Maximum $100 per month


review:
  # Auto-downgrade model if cost limit exceeded
  fallback_model: "gpt-4.1-mini"


  # Skip review if PR is too large
  max_cost_estimate: 1.00        # Skip if estimated cost > $1.00

Usage Tracking

tracking:
  enabled: true
  log_file: ~/.kit/usage.log
  metrics_endpoint: https://your-metrics-server.com/api/usage
  team_id: "engineering-team"

Troubleshooting

Common Issues

1. API Key Issues:

Terminal window
# Test API key
curl -H "Authorization: Bearer sk-ant-your_key" \
  https://api.anthropic.com/v1/messages


# Check environment
echo $KIT_ANTHROPIC_TOKEN

2. Model Availability:

Terminal window
# List available models
kit review --list-models


# Test specific model
kit review --model claude-sonnet-4 --dry-run <pr-url>

3. GitHub Permissions:

Terminal window
# Test GitHub token
curl -H "Authorization: token ghp_your_token" \
  https://api.github.com/user


# Check permissions
gh auth status

Debug Mode

Terminal window
# Enable debug logging
export KIT_DEBUG=true
kit review --dry-run <pr-url>


# Verbose output
kit review --verbose <pr-url>

Configuration Reset

Terminal window
# Reset to defaults
rm ~/.kit/review-config.yaml
kit review --init-config


# Backup current config
cp ~/.kit/review-config.yaml ~/.kit/review-config.yaml.backup

← Back to PR Reviewer Overview