gh-k-dense-ai-claude-scientific-skills-scientific-skills/skills/denario/references/llm_configuration.md

LLM API Configuration

Overview

Denario requires API credentials from supported LLM providers to power its multiagent research system. The system is built on AG2 and LangGraph, which support multiple LLM backends.

Supported LLM Providers

Google Vertex AI

• Full integration with Google's Vertex AI platform
• Supports Gemini and PaLM models
• Requires Google Cloud project setup

OpenAI

• GPT-4, GPT-3.5, and other OpenAI models
• Direct API integration

Other Providers

• Any LLM compatible with AG2/LangGraph frameworks
• Anthropic Claude (via compatible interfaces)
• Azure OpenAI
• Custom model endpoints

Obtaining API Keys

Google Vertex AI

1. Create Google Cloud Project

   • Navigate to Google Cloud Console
   • Create a new project or select existing

2. Enable Vertex AI API

   • Go to "APIs & Services" → "Library"
   • Search for "Vertex AI API"
   • Click "Enable"

3. Create Service Account

   • Navigate to "IAM & Admin" → "Service Accounts"
   • Create service account with Vertex AI permissions
   • Download JSON key file

4. Set up authentication

   export GOOGLE_APPLICATION_CREDENTIALS="/path/to/service-account-key.json"
   

OpenAI

1. Create OpenAI Account

   • Visit platform.openai.com
   • Sign up or log in

2. Generate API Key

   • Navigate to API Keys section
   • Click "Create new secret key"
   • Copy and store securely

3. Set environment variable

   export OPENAI_API_KEY="sk-..."
   

Storing API Keys

Method 1: Environment Variables (Recommended)

Linux/macOS:

export OPENAI_API_KEY="your-key-here"
export GOOGLE_APPLICATION_CREDENTIALS="/path/to/credentials.json"

Add to ~/.bashrc, ~/.zshrc, or ~/.bash_profile for persistence.

Windows:

set OPENAI_API_KEY=your-key-here

Or use System Properties → Environment Variables for persistence.

Method 2: .env Files

Create a .env file in your project directory:

# OpenAI Configuration
OPENAI_API_KEY=sk-your-openai-key-here
OPENAI_MODEL=gpt-4

# Google Vertex AI Configuration
GOOGLE_APPLICATION_CREDENTIALS=/path/to/service-account.json
GOOGLE_CLOUD_PROJECT=your-project-id

# Optional: Model preferences
DEFAULT_MODEL=gpt-4
TEMPERATURE=0.7

Load the environment file in Python:

from dotenv import load_dotenv
load_dotenv()

from denario import Denario
den = Denario(project_dir="./project")

Method 3: Docker Environment Files

For Docker deployments, pass environment variables:

# Using --env-file flag
docker run -p 8501:8501 --env-file .env --rm pablovd/denario:latest

# Using -e flag for individual variables
docker run -p 8501:8501 \
  -e OPENAI_API_KEY=sk-... \
  -e GOOGLE_APPLICATION_CREDENTIALS=/credentials.json \
  -v /local/path/to/creds.json:/credentials.json \
  --rm pablovd/denario:latest

Vertex AI Detailed Setup

Prerequisites

• Google Cloud account with billing enabled
• gcloud CLI installed (optional but recommended)

Step-by-Step Configuration

1. Install Google Cloud SDK (if not using Docker)

   # Linux/macOS
   curl https://sdk.cloud.google.com | bash
   exec -l $SHELL
   gcloud init
   

2. Authenticate gcloud

   gcloud auth application-default login
   

3. Set project

   gcloud config set project YOUR_PROJECT_ID
   

4. Enable required APIs

   gcloud services enable aiplatform.googleapis.com
   gcloud services enable compute.googleapis.com
   

5. Create service account (alternative to gcloud auth)

   gcloud iam service-accounts create denario-service-account \
     --display-name="Denario AI Service Account"
   
   gcloud projects add-iam-policy-binding YOUR_PROJECT_ID \
     --member="serviceAccount:denario-service-account@YOUR_PROJECT_ID.iam.gserviceaccount.com" \
     --role="roles/aiplatform.user"
   
   gcloud iam service-accounts keys create credentials.json \
     --iam-account=denario-service-account@YOUR_PROJECT_ID.iam.gserviceaccount.com
   

6. Configure denario to use Vertex AI

   import os
   os.environ['GOOGLE_CLOUD_PROJECT'] = 'YOUR_PROJECT_ID'
   os.environ['GOOGLE_APPLICATION_CREDENTIALS'] = '/path/to/credentials.json'
   
   from denario import Denario
   den = Denario(project_dir="./research")
   

Model Selection

Configure which models denario uses for different tasks:

# In your code
from denario import Denario

# Example configuration (if supported by denario API)
den = Denario(
    project_dir="./project",
    # Model configuration may vary based on denario version
)

Check denario's documentation for specific model selection APIs.

Cost Management

Monitoring Costs

• OpenAI: Track usage at platform.openai.com/usage
• Google Cloud: Monitor in Cloud Console → Billing
• Set up billing alerts to avoid unexpected charges

Cost Optimization Tips

1. Use appropriate model tiers

   • GPT-3.5 for simpler tasks
   • GPT-4 for complex reasoning

2. Batch operations

   • Process multiple research tasks in single sessions

3. Cache results

   • Reuse generated ideas, methods, and results when possible

4. Set token limits

   • Configure maximum token usage for cost control

Security Best Practices

Do NOT commit API keys to version control

Add to .gitignore:

.env
*.json  # If storing credentials
credentials.json
service-account-key.json

Rotate keys regularly

• Generate new API keys periodically
• Revoke old keys after rotation

Use least privilege access

• Grant only necessary permissions to service accounts
• Use separate keys for development and production

Encrypt sensitive files

• Store credential files in encrypted volumes
• Use cloud secret management services for production

Troubleshooting

"API key not found" errors

• Verify environment variables are set: echo $OPENAI_API_KEY
• Check .env file is in correct directory
• Ensure load_dotenv() is called before importing denario

Vertex AI authentication failures

• Verify GOOGLE_APPLICATION_CREDENTIALS points to valid JSON file
• Check service account has required permissions
• Ensure APIs are enabled in Google Cloud project

Rate limiting issues

• Implement exponential backoff
• Reduce concurrent requests
• Upgrade API plan if needed

Docker environment variable issues

• Use docker run --env-file .env to pass environment
• Mount credential files with -v flag
• Check environment inside container: docker exec <container> env