Agent Integrations

R9S works with tools that allow a custom OpenAI-compatible base URL. The integrations below are listed only after end-to-end verification with a supported R9S model.

Common OpenAI-compatible base URL:

https://gw.r9s.ai/v1

Use your R9S API key for authentication.

Verify access

Before configuring an agent, confirm that the key can list models:

export R9S_API_KEY="<R9S_API_KEY>"

curl https://gw.r9s.ai/v1/models \
  -H "Authorization: Bearer $R9S_API_KEY"

Supported patterns

Codex

Configure Codex with an OpenAI-compatible base URL:

export R9S_API_KEY="<R9S_API_KEY>"
export OPENAI_BASE_URL="https://gw.r9s.ai/v1"
export OPENAI_API_KEY="$R9S_API_KEY"

Run with a supported R9S model:

codex --model glm-5.3

OpenCode

Add an R9S provider to OpenCode configuration:

{
  "$schema": "https://opencode.ai/config.json",
  "provider": {
    "r9s": {
      "npm": "@ai-sdk/openai-compatible",
      "name": "R9S",
      "options": {
        "baseURL": "https://gw.r9s.ai/v1"
      },
      "models": {
        "glm-5.3": {
          "name": "GLM-5.3"
        }
      }
    }
  }
}

Launch OpenCode, run /connect, choose R9S, and paste the R9S API key. Then use /models to select an R9S-backed model.

OpenClaw

OpenClaw can use a LiteLLM-style custom base URL:

export R9S_API_KEY="<R9S_API_KEY>"
export LITELLM_API_KEY="$R9S_API_KEY"

Onboard with R9S:

openclaw onboard \
  --non-interactive \
  --accept-risk \
  --auth-choice litellm-api-key \
  --litellm-api-key "$R9S_API_KEY" \
  --custom-base-url "https://gw.r9s.ai/v1" \
  --install-daemon \
  --skip-channels \
  --skip-skills

Select a model routed through R9S:

openclaw models set litellm/glm-5.3

Troubleshooting

Operational notes

• Prefer model names that have been verified through /v1/models.
• Keep agent keys scoped and rotated separately from application keys.
• Use separate API keys per agent team when you need usage attribution.
• Watch cost and retry behavior before giving agents high autonomy.