Configuration

Core Environment Variables

• OPENCLAW_IMAGE: OpenClaw image to run
• OPENCLAW_PODMAN_PUBLISH_HOST: host publish address
• OPENCLAW_OLLAMA_BASE_URL: Ollama native API URL
• OPENCLAW_OLLAMA_MODEL: default Ollama model id
• OPENCLAW_MODEL_REF: provider-qualified model reference such as zai/glm-5.1, ollama/gemma4:e2b, google/gemma-4-31b-it, or nvidia/moonshotai/kimi-k2.6
• OPENCLAW_MODEL_FALLBACKS: comma-separated fallback model refs; leave it empty for providers that should not inherit Z.AI fallbacks
• OPENCLAW_ZAI_BASE_URL, OPENCLAW_OPENROUTER_BASE_URL, OPENCLAW_NVIDIA_BASE_URL: provider endpoints used when the selected model ref targets those providers
• OPENCLAW_SCALE_INSTANCE_ROOT: root directory for generated instances
• OPENCLAW_SCALE_GATEWAY_PORT_START: first gateway port
• OPENCLAW_SCALE_BRIDGE_PORT_START: first bridge port
• OPENCLAW_SCALE_PORT_STEP: per-instance port increment

Provider API keys stay in .env and are injected by provider name. The currently supported keys are ZAI_API_KEY, OPENROUTER_API_KEY, GEMINI_API_KEY, and NVIDIA_API_KEY.

Agent-Team Tuning

These env vars control the communication lab:

• OPENCLAW_MATTERMOST_ENABLED
• OPENCLAW_MATTERMOST_CHANNEL_NAME
• OPENCLAW_MATTERMOST_AUTONOMY_ENABLED
• OPENCLAW_MATTERMOST_AUTONOMY_INTERVAL
• OPENCLAW_MATTERMOST_AUTONOMY_INTERVAL_INSTANCE_00N
• OPENCLAW_MATTERMOST_AUTONOMY_MODEL
• OPENCLAW_MODEL_REF_INSTANCE_00N
• OPENCLAW_MODEL_FALLBACKS_INSTANCE_00N
• OPENCLAW_MATTERMOST_AUTONOMY_MODEL_INSTANCE_00N

Most team behavior still lives in the workspace scaffolds rather than env vars. Use .env for infrastructure defaults, then tune SOUL.md, IDENTITY.md, USER.md, and HEARTBEAT.md inside each generated workspace.

Per-instance model overrides let the same team mix providers. .env.example includes examples for Google/Gemma seats 4-6 and optional NVIDIA seats 7-9; use those with --count 9 only when the matching provider keys are available.

Important Note For Windows + WSL Podman

The working Ollama endpoint observed during validation was:

http://172.27.208.1:11434

If host.containers.internal does not reach the Windows-hosted Ollama instance in your environment, replace OPENCLAW_OLLAMA_BASE_URL accordingly.

Trust Model

This project targets same-trust operator workflows. OpenClaw is configured in a full-access-in-container mode and depends on the Podman pod boundary rather than OpenClaw's internal sandbox for the main isolation layer.