Environment Variables

llmist uses environment variables for provider authentication and configuration.

Provider API Keys

Set these environment variables to enable provider auto-discovery:

# OpenAI
export OPENAI_API_KEY="sk-..."

# Anthropic
export ANTHROPIC_API_KEY="sk-ant-..."

# Google Gemini
export GEMINI_API_KEY="..."

# OpenRouter (400+ models via unified gateway)
export OPENROUTER_API_KEY="sk-or-..."

# HuggingFace (free tier, serverless inference)
export HF_TOKEN="hf_..."

Tip

llmist automatically discovers and configures providers based on which API keys are set. No manual provider registration required.

Auto-Discovery Behavior

When you create an agent without specifying a provider, llmist:

1. Checks for API keys in environment variables
2. Registers providers with valid keys
3. Selects the appropriate provider based on model name

// Auto-discovers providers from environment
const agent = LLMist.createAgent()
  .withModel('sonnet');  // Uses Anthropic if ANTHROPIC_API_KEY is set

Logging Configuration

Variable	Description	Values
LLMIST_LOG_LEVEL	Logging verbosity	silly, trace, debug, info, warn, error, fatal
LLMIST_LOG_FILE	Write logs to a file instead of console	File path
LLMIST_LOG_RESET	Truncate log file on startup	true, false
LLMIST_LOG_TEE	Also write to stdout when file logging is active	true, false

Example Logging Setup

# Enable debug logging
export LLMIST_LOG_LEVEL="debug"

# Log to file AND stdout (useful for Docker/containers)
export LLMIST_LOG_FILE="/tmp/agent.log"
export LLMIST_LOG_TEE="true"

Note

Each CLI session automatically creates a log directory at ~/.llmist/logs/<session-name>/ with a memorable name like sunny-falcon. Numbers are only appended when there’s a name collision (e.g., sunny-falcon-2). Logs are stored in session.log.jsonl within that directory.

CLI-Specific Flags

These can also be set via command-line flags:

Flag	Environment Variable	Description
--log-level	LLMIST_LOG_LEVEL	Set log level

    # Via flag
    npx @llmist/cli agent "Hello" --log-level debug

    # Via environment
    LLMIST_LOG_LEVEL=debug npx @llmist/cli agent "Hello"

Provider-Specific Variables

OpenAI

Variable	Description
OPENAI_API_KEY	API key (required)
OPENAI_ORG_ID	Organization ID (optional)
OPENAI_BASE_URL	Custom API endpoint (optional)

Anthropic

Variable	Description
ANTHROPIC_API_KEY	API key (required)
ANTHROPIC_BASE_URL	Custom API endpoint (optional)

Google Gemini

Variable	Description
GEMINI_API_KEY	API key (required)

OpenRouter

Variable	Description
OPENROUTER_API_KEY	API key (required)
OPENROUTER_SITE_URL	Your app URL for analytics (optional)
OPENROUTER_APP_NAME	Your app name for analytics (optional)

HuggingFace

Access thousands of open-source models via HuggingFace’s serverless inference router.

Variable	Description
HF_TOKEN	HuggingFace access token (primary, required)
HUGGING_FACE_API_KEY	Fallback token if HF_TOKEN is not set (optional)
HF_ENDPOINT_URL	Custom inference endpoint URL (optional, for dedicated endpoints)

# Standard serverless inference (router.huggingface.co)
export HF_TOKEN="hf_..."

# Dedicated endpoint (overrides the default router URL)
export HF_TOKEN="hf_..."
export HF_ENDPOINT_URL="https://xxx.endpoints.huggingface.cloud"

Tip

Get a free HuggingFace token at huggingface.co/settings/tokens. The free tier provides access to thousands of open-source models including Llama, Mistral, Qwen, and DeepSeek via huggingface: or hf: prefix.

Best Practices

1. Never commit API keys - Use .env files or secret managers
2. Use different keys per environment - Separate dev/staging/production
3. Rotate keys regularly - Update keys periodically for security

Using .env Files

# .env (add to .gitignore!)
OPENAI_API_KEY=sk-...
ANTHROPIC_API_KEY=sk-ant-...

Load with your preferred method:

• Node.js: dotenv package
• Bun: Built-in .env support
• Shell: source .env or export $(cat .env | xargs)