EPAMEPAMDMToolsDocs
GitHubInstall DMTools

Google Gemini AI Configuration

Overview

Google Gemini provides powerful AI capabilities with a generous free tier (15 requests/minute), making it ideal for getting started with DMtools. Gemini 2.0 Flash offers excellent performance for most tasks.

Quick Setup

Step 1: Get API Key

  1. Go to Google AI Studio
  2. Click “Get API Key”
  3. Select or create a Google Cloud project
  4. Click “Create API key in existing project”
  5. Copy your API key (starts with AIzaSy...)

Step 2: Configure DMtools

Add to your dmtools.env:

# Gemini Configuration
GEMINI_API_KEY=YOUR_GEMINI_API_KEY_HERE
GEMINI_MODEL=gemini-2.0-flash-exp  # Optional, defaults to gemini-2.0-flash-exp

Step 3: Test Configuration

# Test Gemini connection
dmtools gemini_ai_chat "Hello, please confirm you're working"

# Expected response:
# "Hello! Yes, I'm working properly. I'm Gemini, ready to help you with your tasks."

Available Models

GEMINI_MODEL=gemini-2.0-flash-exp
# - Fastest response time
# - Best for most DMtools tasks
# - 1M token context window
# - Free tier: 15 RPM, 1M TPM, 1500 RPD

Gemini 1.5 Pro

GEMINI_MODEL=gemini-1.5-pro-002
# - More capable for complex tasks
# - 2M token context window
# - Free tier: 2 RPM, 32K TPM, 50 RPD
# - Use for large code analysis

Gemini 1.5 Flash

GEMINI_MODEL=gemini-1.5-flash-002
# - Previous generation flash model
# - Good balance of speed and capability
# - Free tier: 15 RPM, 1M TPM, 1500 RPD

Advanced Configuration

Rate Limit Handling

# dmtools.env
GEMINI_API_KEY=YOUR_GEMINI_API_KEY_HERE
GEMINI_MODEL=gemini-2.0-flash-exp
GEMINI_RETRY_ATTEMPTS=3          # Retry on rate limit
GEMINI_RETRY_DELAY_MS=2000       # Wait 2 seconds between retries

Multiple API Keys (Load Balancing)

# For higher throughput, rotate between keys
GEMINI_API_KEY_1=YOUR_FIRST_GEMINI_API_KEY
GEMINI_API_KEY_2=YOUR_SECOND_GEMINI_API_KEY
GEMINI_API_KEY_3=YOUR_THIRD_GEMINI_API_KEY
GEMINI_LOAD_BALANCE=true

Context Window Management

# Control context size for large operations
PROMPT_CHUNK_TOKEN_LIMIT=4000    # Max tokens per chunk
PROMPT_CHUNK_MAX_SINGLE_FILE_SIZE_MB=4  # Max file size

Usage Examples

Example 1: Test Case Generation

# Generate test cases from Jira story
dmtools TestCasesGenerator --inputJql "key = PROJ-123"

# Uses Gemini to analyze story and create test scenarios

Example 2: Code Analysis

// agents/js/codeReview.js
function action(params) {
    const code = file_read(params.filePath);

    const analysis = gemini_ai_chat(`
        Analyze this code for:
        1. Security vulnerabilities
        2. Performance issues
        3. Code quality

        Code:
        ${code}
    `);

    return JSON.parse(analysis);
}

Example 3: Documentation Generation

// agents/doc_generator.json
{
  "name": "DocumentationGenerator",
  "params": {
    "aiProvider": "gemini",
    "aiModel": "gemini-2.0-flash-exp",
    "aiRole": "You are a technical documentation expert",
    "instructions": "Generate comprehensive API documentation"
  }
}

Example 4: Direct AI Chat

# Interactive chat
dmtools gemini_ai_chat "Explain the SOLID principles with examples"

# From file
echo "Review this SQL query for optimization" > prompt.txt
dmtools gemini_ai_chat --file prompt.txt

# With context
dmtools gemini_ai_chat "Analyze ticket PROJ-123" --context "$(dmtools jira_get_ticket PROJ-123)"

Optimizing for Gemini

1. Prompt Engineering

// Good prompt structure for Gemini
const prompt = `
Role: You are an expert QA engineer.

Context: ${ticketDescription}

Task: Generate comprehensive test cases.

Requirements:
1. Cover positive and negative scenarios
2. Include edge cases
3. Follow Given-When-Then format

Output Format: JSON array with test cases
`;

2. Token Optimization

# Monitor token usage
dmtools --debug gemini_ai_chat "test prompt"
# Shows: Tokens used: input=50, output=200

# Reduce context for faster responses
PROMPT_CHUNK_TOKEN_LIMIT=2000  # Smaller chunks

Troubleshooting

API Key Issues

# Error: "API key not valid"

# Verify key format
echo $GEMINI_API_KEY
# Should start with AIzaSy...

# Test directly with curl
curl -X POST "https://generativelanguage.googleapis.com/v1/models/gemini-2.0-flash:generateContent?key=$GEMINI_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"contents":[{"parts":[{"text":"Hello"}]}]}'

Context Length Exceeded

# Error: "Request too large"

# Reduce context size
PROMPT_CHUNK_TOKEN_LIMIT=2000
PROMPT_CHUNK_MAX_SINGLE_FILE_SIZE_MB=2

# Or use model with larger context
GEMINI_MODEL=gemini-1.5-pro-002  # 2M tokens

Timeout Issues

# Error: "Read timed out"

# Increase timeout
export GEMINI_TIMEOUT_SECONDS=60

# Or use streaming for long responses
GEMINI_STREAM_RESPONSE=true

Security Best Practices

1. Secure API Key Storage

# Never commit API keys
echo "*.env" >> .gitignore

# Use environment variables in CI/CD
export GEMINI_API_KEY=${{ secrets.GEMINI_API_KEY }}

2. Restrict API Key Usage

In Google Cloud Console:

  1. Select your API key
  2. Click “Edit API key”
  3. Under “API restrictions”, select “Restrict key”
  4. Choose “Gemini API” only
  5. Add IP restrictions if needed

3. Monitor Usage

# Check daily usage
curl "https://generativelanguage.googleapis.com/v1/models?key=$GEMINI_API_KEY"

# Set up alerts in Google Cloud Console for unusual activity

Best Practices

1. Start with Free Tier

  • Perfect for development and testing
  • 15 requests/minute is sufficient for most workflows
  • No credit card required

2. Use Flash for Speed

  • gemini-2.0-flash-exp is 2-3x faster
  • Ideal for iterative development
  • Handles structured output well

3. Cache Responses

// Cache AI responses to avoid repeated calls
const cache = {};
function getCachedResponse(prompt) {
    const key = hash(prompt);
    if (cache[key]) return cache[key];

    const response = gemini_ai_chat(prompt);
    cache[key] = response;
    return response;
}

4. Batch Operations

// Process multiple items in one request
const items = ["item1", "item2", "item3"];
const response = gemini_ai_chat(`
    Process these items:
    ${items.join('\n')}

    Return as JSON array.
`);

Integration with DMtools Jobs

Configure for Test Generation

{
  "name": "TestCasesGenerator",
  "params": {
    "aiProvider": "gemini",
    "aiModel": "gemini-2.0-flash-exp",
    "inputJql": "sprint in openSprints()"
  }
}

Useful Resources

Next: OpenAI Configuration | Azure DevOps Setup

Edit this page on GitHub →

Rendered from dmtools-ai-docs/references/configuration/ai-providers/gemini.md, the same Markdown the DMTools agent skill reads. Last updated .