Skip to main content
Forge supports multiple AI providers. Use the interactive CLI to manage your provider credentials securely.

​
Quick Start

# Login to a provider
forge provider login

# List supported providers
forge provider list

# Remove provider credentials
forge provider logout
On first run, Forge automatically guides you through provider setup if no credentials are found.

​
Provider Management

​
Interactive Login

The recommended way to configure providers: 
forge provider login
This command:
  1. Shows available providers
  2. Guides you through credential entry
  3. Securely stores credentials in file-based storage
  4. Validates the credentials

​
Listing Providers

View all supported providers: 
forge provider list

​
Removing Credentials

Remove stored credentials: 
forge provider logout

​
Supported Providers

Claude models from Anthropic.Setup:
forge provider login
# Select Anthropic and enter your API key
Configure model:
# forge.yaml
model: claude-3.7-sonnet
Available models:
  • claude-3.7-sonnet - Latest Sonnet model
  • claude-sonnet-4 - Claude Sonnet 4
  • claude-3-opus - Most capable model
  • claude-3-haiku - Fastest model
GPT models from OpenAI.Setup:
forge provider login
# Select OpenAI and enter your API key
Configure model:
# forge.yaml
model: o3-mini-high
Available models:
  • o3-mini-high - High reasoning capability
  • gpt-4-turbo - Advanced GPT-4 model
  • gpt-4 - Standard GPT-4
Gemini and Claude models via Google Cloud.Setup:
  1. Install and authenticate with Google Cloud CLI:
gcloud auth login
gcloud config set project YOUR_PROJECT_ID
  1. Get authentication token:
gcloud auth print-access-token
  1. Configure in Forge:
forge provider login
# Select Google Vertex AI and enter credentials
Configure model:
# forge.yaml
model: google/gemini-2.5-pro
Available models:
  • gemini-2.5-pro - Latest Gemini Pro
  • gemini-2.0-flash - Fast Gemini model
  • claude-sonnet-4@20250514 - Claude via Vertex AI
Access multiple AI models through a single API.Setup:
forge provider login
# Select OpenRouter and enter your API key
Grok models from x.ai.Setup:
forge provider login
# Select x.ai and enter your API key
Fast inference models from Cerebras.Setup:
forge provider login
# Select Cerebras and enter your API key
Claude and other models via AWS Bedrock.Requirements:
  • Deploy Bedrock Access Gateway
  • Create API key in AWS Secrets Manager
  • Note the API Base URL from CloudFormation outputs
Setup:
forge provider login
# Select OpenAI-compatible provider
# Enter Bedrock Gateway URL and API key
Configure model:
# forge.yaml
model: anthropic.claude-3-opus
Fast inference using Groq hardware.Setup:
forge provider login
# Select OpenAI-compatible provider
# Enter URL: https://api.groq.com/openai/v1
# Enter your Groq API key
Configure model:
# forge.yaml
model: deepseek-r1-distill-llama-70b
Any provider with OpenAI-compatible API.Setup:
forge provider login
# Select OpenAI-compatible provider
# Enter provider URL and API key
Configure model:
# forge.yaml
model: <provider-specific-model>
Managed AI services from Forge.Setup:
forge provider login
# Select Forge Services and enter your API key
Configure model:
# forge.yaml
model: claude-3.7-sonnet

​
Environment Variables (Deprecated)

Using environment variables for provider configuration is deprecated and will be removed in a future version. Use forge provider login instead.
For backward compatibility, Forge still supports environment variables. Credentials found in environment variables are automatically migrated to file-based storage on first run. 
# .env (deprecated)
OPENAI_API_KEY=<your_openai_api_key>

​
Model Selection

After configuring a provider, specify the model in your forge.yaml: 
# forge.yaml
model: claude-3.7-sonnet
Or use the /model command in the Forge CLI to see all available models for your configured providers.

​
Advanced Configuration

​
Custom API Endpoints

Override default API URLs using environment variables: 
# .env
FORGE_API_URL=https://api.forgecode.dev
FORGE_WORKSPACE_SERVER_URL=http://localhost:8080

​
HTTP Client Configuration

Fine-tune API request behavior: 
# .env
FORGE_HTTP_CONNECT_TIMEOUT=30
FORGE_HTTP_READ_TIMEOUT=900
FORGE_HTTP_MAX_REDIRECTS=10
See the environment variables reference for all HTTP configuration options.

​
Troubleshooting

If you see authentication errors:
  1. Run forge provider logout to remove old credentials
  2. Run forge provider login to re-enter credentials
  3. Verify your API key is valid on the provider’s website
If you hit rate limits, configure retry behavior:
# .env
FORGE_RETRY_MAX_ATTEMPTS=5
FORGE_RETRY_INITIAL_BACKOFF_MS=2000
FORGE_RETRY_BACKOFF_FACTOR=2
For slow connections, increase timeout values:
# .env
FORGE_HTTP_CONNECT_TIMEOUT=60
FORGE_HTTP_READ_TIMEOUT=1800
If you encounter SSL certificate issues:
# .env
# Add custom certificate paths
FORGE_HTTP_ROOT_CERT_PATHS=/path/to/cert1.pem,/path/to/cert2.crt
Warning: Only use FORGE_HTTP_ACCEPT_INVALID_CERTS=true in development.

​
Security Best Practices

Credential Storage

  • Never commit API keys to version control
  • Use forge provider login for secure credential storage
  • Credentials are stored in platform-specific secure locations
  • Environment variables are auto-migrated on first run

​
Next Steps

forge.yaml Configuration

Configure model behavior and workflow settings

Custom Rules

Define project-specific guidelines

Build docs developers (and LLMs) love

Get started for freeTalk to us