Skip to content

Init Command Reference

Home › Reference › Commands › Other › Init

Overview

The init command creates the AI provider configuration file at .r2r/eac/ai-provider.yml. This configuration enables AI-powered features throughout EAC including commit message generation, specification creation, architecture design, and pull request descriptions.

AI Provider Comparison

Provider API Key Required Cost Best For Model
claude-api ANTHROPIC_API_KEY Pay-per-use Production use, high volume Claude Sonnet 4.5
claude-cli None (Claude Pro subscription) Included with subscription Development, lower volume Claude Sonnet 4.5
openai OPENAI_API_KEY Pay-per-use OpenAI ecosystem integration GPT-4
gemini GOOGLE_API_KEY Pay-per-use Google Cloud integration Gemini Pro

Provider Selection Guide

Choose claude-api when:

  • You need production-grade reliability
  • You have high-volume AI operations
  • You want the latest Claude models
  • You're willing to pay per API call

Choose claude-cli when:

  • You have a Claude Pro subscription
  • You're doing development work
  • You have moderate AI usage needs
  • You want to avoid API costs

Choose openai when:

  • You're already using OpenAI services
  • You prefer GPT-4 models
  • You have existing OpenAI infrastructure

Choose gemini when:

  • You're using Google Cloud Platform
  • You prefer Google's AI models
  • You have Google Cloud credits

Cost Optimization

For claude-cli:

  • Free usage included with Claude Pro subscription
  • Rate limits apply
  • Best for development work

For claude-api, openai, gemini:

  • Monitor usage through provider dashboard
  • Set spending limits in provider console
  • Use caching when available
  • Start with smaller models for testing

Configuration File Format

The init command creates .r2r/eac/ai-provider.yml with provider-specific configuration.

Claude API Configuration

provider: claude-api
api_key_env: ANTHROPIC_API_KEY
model: claude-sonnet-4-5
max_tokens: 4096
temperature: 0.7

Configuration Options:

  • provider: Must be claude-api
  • api_key_env: Environment variable name containing the API key
  • model: Claude model to use (default: claude-sonnet-4-5)
  • max_tokens: Maximum tokens in response (default: 4096)
  • temperature: Randomness in responses, 0.0-1.0 (default: 0.7)

Claude CLI Configuration

provider: claude-cli
model: claude-sonnet-4-5

Configuration Options:

  • provider: Must be claude-cli
  • model: Claude model to use (default: claude-sonnet-4-5)

Note: No API key required. Uses Claude Pro subscription authentication.

OpenAI Configuration

provider: openai
api_key_env: OPENAI_API_KEY
model: gpt-4
max_tokens: 4096
temperature: 0.7

Configuration Options:

  • provider: Must be openai
  • api_key_env: Environment variable name containing the API key
  • model: OpenAI model to use (default: gpt-4)
  • max_tokens: Maximum tokens in response (default: 4096)
  • temperature: Randomness in responses, 0.0-1.0 (default: 0.7)

Gemini Configuration

provider: gemini
api_key_env: GOOGLE_API_KEY
model: gemini-pro
max_tokens: 4096
temperature: 0.7

Configuration Options:

  • provider: Must be gemini
  • api_key_env: Environment variable name containing the API key
  • model: Gemini model to use (default: gemini-pro)
  • max_tokens: Maximum tokens in response (default: 4096)
  • temperature: Randomness in responses, 0.0-1.0 (default: 0.7)

Environment Variable Setup

Windows (PowerShell)

# Temporary (current session only)
$env:ANTHROPIC_API_KEY = "sk-ant-api03-..."
$env:OPENAI_API_KEY = "sk-proj-..."
$env:GOOGLE_API_KEY = "AIza..."

# Permanent (user environment variable)
[System.Environment]::SetEnvironmentVariable('ANTHROPIC_API_KEY', 'sk-ant-api03-...', 'User')
[System.Environment]::SetEnvironmentVariable('OPENAI_API_KEY', 'sk-proj-...', 'User')
[System.Environment]::SetEnvironmentVariable('GOOGLE_API_KEY', 'AIza...', 'User')

Windows (Command Prompt)

# Temporary (current session only)
set ANTHROPIC_API_KEY=sk-ant-api03-...
set OPENAI_API_KEY=sk-proj-...
set GOOGLE_API_KEY=AIza...

# Permanent (requires admin)
setx ANTHROPIC_API_KEY "sk-ant-api03-..."
setx OPENAI_API_KEY "sk-proj-..."
setx GOOGLE_API_KEY "AIza..."

Linux/macOS (Bash/Zsh)

# Temporary (current session only)
export ANTHROPIC_API_KEY="sk-ant-api03-..."
export OPENAI_API_KEY="sk-proj-..."
export GOOGLE_API_KEY="AIza..."

# Permanent (add to ~/.bashrc or ~/.zshrc)
echo 'export ANTHROPIC_API_KEY="sk-ant-api03-..."' >> ~/.bashrc
echo 'export OPENAI_API_KEY="sk-proj-..."' >> ~/.bashrc
echo 'export GOOGLE_API_KEY="AIza..."' >> ~/.bashrc

# Reload configuration
source ~/.bashrc

Verify Environment Variables

# Check if variable is set
echo $ANTHROPIC_API_KEY
echo $OPENAI_API_KEY
echo $GOOGLE_API_KEY

# Should display your API key
# If empty, the variable is not set

CI/CD Setup

# Store API key in CI/CD secrets
# GitHub Actions example:
# Add ANTHROPIC_API_KEY to repository secrets

# In your workflow:
- name: Initialize AI provider
  env:
    ANTHROPIC_API_KEY: ⟪ secrets.ANTHROPIC_API_KEY ⟫
  run: r2r eac init --ai claude-api

Configuration File Location

The configuration file is stored at:

.r2r/eac/ai-provider.yml

Important: This file should NOT be committed to version control as it may contain sensitive information or preferences specific to your environment.

Recommended .gitignore Entry

# AI provider configuration (personal settings)
.r2r/eac/ai-provider.yml

Troubleshooting

Problem: API Key Not Found

Error:

Error: ANTHROPIC_API_KEY environment variable not set

Solution:

# Verify variable is set
echo $ANTHROPIC_API_KEY

# If empty, set it
export ANTHROPIC_API_KEY="sk-ant-api03-..."

# Try again
r2r eac init --ai claude-api

Problem: Invalid API Key

Error:

Error: Authentication failed - invalid API key

Solution:

  1. Verify you copied the complete key (no spaces)
  2. Check key hasn't expired
  3. Confirm key is for the correct provider
  4. Generate a new key from provider dashboard

Problem: Configuration File Already Exists

Error:

Warning: Configuration file already exists at .r2r/eac/ai-provider.yml

Solution:

# The command will overwrite the existing file
# Your choice:
# 1. Proceed (overwrites old config)
r2r eac init --ai claude-api

# 2. Back up old config first
cp .r2r/eac/ai-provider.yml .r2r/eac/ai-provider.yml.backup
r2r eac init --ai claude-api

Problem: Permission Denied

Error:

Error: Permission denied: .r2r/eac/ai-provider.yml

Solution:

# Check directory permissions
ls -la .r2r/eac/

# Fix permissions
chmod 755 .r2r/eac
chmod 644 .r2r/eac/ai-provider.yml

# Try again
r2r eac init --ai claude-api

Problem: Provider Not Recognized

Error:

Error: Unknown provider: claude

Solution:

# Use exact provider name
# Valid options: claude-api, claude-cli, openai, gemini
r2r eac init --ai claude-api  # ✓ Correct
r2r eac init --ai claude      # ✗ Invalid

Problem: AI Features Not Working After Init

Symptoms:

  • r2r eac work commit doesn't generate messages
  • r2r eac create spec fails
  • No AI responses

Solution:

# 1. Verify configuration exists
cat .r2r/eac/ai-provider.yml

# 2. Check environment variable
echo $ANTHROPIC_API_KEY  # or appropriate variable

# 3. Test with debug mode
r2r eac work commit --all --debug

# 4. Re-initialize if needed
r2r eac init --ai claude-api --debug

Technical Notes

  • The init command can be run multiple times to change providers
  • Each run overwrites the previous configuration
  • Environment variables are checked during initialization
  • API keys are never stored in the configuration file (only the environment variable name)
  • The command validates the provider name but does not validate API keys
  • Configuration is local to each repository
  • Team members can use different providers if desired
  • The .r2r/eac/ directory is created automatically if it doesn't exist
  • work commit - Generate AI-powered commit messages
  • create spec - Generate Gherkin specifications with AI
  • create design - Generate architecture diagrams with AI
  • work pr - Create pull requests with AI-generated descriptions
  • show config - Display current configuration

See Also

Tutorials | How-to Guides | Explanation | Reference

You are here: Reference — information-oriented technical descriptions of the system.