Skip to main content

Configure LLM

Cognitive Modules (Node CLI) supports multiple LLM providers. The CLI auto-selects a provider based on which API key is present, or you can force one with --provider.

Supported Providers​

ProviderEnvironment Variable
OpenAIOPENAI_API_KEY
AnthropicANTHROPIC_API_KEY
GeminiGEMINI_API_KEY
DeepSeekDEEPSEEK_API_KEY
MiniMaxMINIMAX_API_KEY
Moonshot (Kimi)MOONSHOT_API_KEY
Qwen (DashScope)DASHSCOPE_API_KEY or QWEN_API_KEY
Ollama (local)OLLAMA_HOST

Example: OpenAI​

export OPENAI_API_KEY=sk-xxx
npx cogn@2.2.12 run code-reviewer --args "your code" --provider openai --model gpt-4o

Example: Anthropic​

export ANTHROPIC_API_KEY=sk-ant-xxx
npx cogn@2.2.12 run code-reviewer --args "your code" --provider anthropic --model claude-sonnet-4.5

Example: Ollama (Local)​

# Install and start Ollama separately
export OLLAMA_HOST=http://localhost:11434
npx cogn@2.2.12 run code-reviewer --args "your code" --provider ollama --model llama3.1

Model Override​

  • CLI: --model <name>
  • Env: COG_MODEL (global override)

Check Configuration​

npx cogn@2.2.12 doctor