Skip to content

connect-providers.md

Latest commit

 

History

History
54 lines (36 loc) · 1.47 KB

connect-providers.md

File metadata and controls

54 lines (36 loc) · 1.47 KB
Raw

Connect AI Providers

CrewCmd supports multiple AI providers. Mix and match models across your agent team.

Supported Providers

Provider Models Best For
Anthropic Claude Opus, Sonnet, Haiku Coding, analysis, complex reasoning
OpenAI GPT-4o, o1, o3 General purpose, fast responses
Google Gemini Pro, Flash Long context, multimodal
Local (Ollama) Llama, Mistral, etc. Privacy, no API costs

Adding API Keys

  1. Go to Settings > Provider Keys
  2. Click Add Key
  3. Select the provider
  4. Paste your API key
  5. Click Save

Keys are encrypted at rest. Each key can be scoped to specific agents or available team-wide.

Connecting Runtimes

For agents that need to execute code or interact with tools:

OpenClaw

Connect agents running on OpenClaw nodes:

  1. Go to Settings > Runtimes
  2. Click Add Runtime
  3. Enter your OpenClaw gateway URL
  4. Agents on that node become available in CrewCmd

Direct API

Connect any agent that exposes an API:

  1. Go to Settings > Runtimes
  2. Click Add Runtime
  3. Configure the endpoint URL and authentication
  4. Map the runtime to specific agents

Per-Agent Model Selection

Each agent can use a different model:

  1. Go to the agent's settings
  2. Under Model, select from available providers
  3. Save

This lets you optimize cost and capability: use a powerful model for complex coding, a faster model for triage, and a local model for sensitive data.