Use OpenClaw with ORGN Gateway
Configure OpenClaw to route agent requests through ORGN Gateway's OpenAI-compatible API — TEE or ZDR models per your selection.
Configure OpenClaw to use ORGN Gateway as its inference backend. Agent messages route to https://api.gateway.orgn.com/v1.
OpenClaw config uses providers.ollm and ollm/<model-id> routing keys — these are OpenClaw integration identifiers, not the product name. Inference is served by ORGN Gateway.
What is OpenClaw?
OpenClaw is an open-source agent platform for Telegram, Discord, Slack, Signal, iMessage, and WhatsApp. It supports custom OpenAI-compatible providers.
Step 1: API key
- Open the Gateway console.
- Create an API key under API Keys.
- Export it:
export OLLM_API_KEY="sk-ollm-YOUR_API_KEY"Step 2: Configure openclaw.json
Edit ~/.openclaw/openclaw.json:
{
"env": {
"OLLM_API_KEY": "your-api-key"
},
"models": {
"mode": "merge",
"providers": {
"ollm": {
"baseUrl": "https://api.gateway.orgn.com/v1",
"apiKey": "${OLLM_API_KEY}",
"api": "openai-completions",
"models": [
{ "id": "near_glm_4_7", "name": "GLM 4.7" },
{ "id": "near_glm_5", "name": "GLM 5" }
]
}
}
},
"agents": {
"defaults": {
"model": {
"primary": "ollm/near_glm_4_7"
},
"models": {
"ollm/near_glm_4_7": {}
}
}
}
}Model IDs must use underscores (near_glm_4_7) — match GET /v1/models.
Step 3: Restart
openclaw gateway restartModel format
OpenClaw references models as:
ollm/<gateway-model-id>Example: ollm/near_glm_4_7
Fallbacks
"model": {
"primary": "ollm/near_glm_4_7",
"fallbacks": ["ollm/near_glm_5"]
}Auth profiles (recommended)
openclaw auth set ollm:default --key "$OLLM_API_KEY""providers": {
"ollm": {
"apiKey": "auth:ollm:default"
}
}Monitoring
Track usage in the Gateway console — requests, tokens, attestation status for TEE models.
Common errors
| Error | Fix |
|---|---|
No API key found for provider 'ollm' | Set OLLM_API_KEY or auth profile |
| 401 | Valid key; baseUrl = https://api.gateway.orgn.com/v1 |
| Model not found | Underscore ID in providers.ollm.models and ollm/<id> reference |