KaibanJS Team as an OpenClaw Agent via the OpenResponses API

Default: Content Creation Team

A sequential 3-agent pipeline that turns any user topic into a researched, written, and reviewed piece of content:

1. ResearchBot — Research Specialist: gathers and analyzes information about the given topic using the configured LLM
2. WriterBot — Content Writer: transforms research findings into polished, engaging content
3. ReviewBot — Quality Reviewer: validates the output against quality standards and refines the final result

The user's message is injected as { topic: userMessage } into the team inputs. Powered by OPENAI_API_KEY.

Bring Your Own KaibanJS Team

The adapter is intentionally decoupled from the team definition. To use a different workflow:

• 1.Open src/team/index.ts and replace the exported createTeam factory with your own KaibanJS Team definition
• 2.Ensure your team accepts the relevant input(s) injected from the user message. Update the extractUserMessage mapping in src/adapter.ts if you need to reshape the input
• 3.Add any additional environment variables your team requires to .env and restart the adapter — no other changes needed

Model Provider Registration

Add the following block under models.providers in your OpenClaw config. Use models.mode: "merge" to keep existing providers intact:

{
  models: {
    mode: "merge",
    providers: {
      "kaiban-adapter": {
        baseUrl: "http://localhost:3100/v1",
        apiKey: "${KAIBAN_OPENRESPONSES_SECRET}",
        api: "openai-responses",
        models: [{
          id: "kaiban",
          name: "KaibanJS Team",
          reasoning: false,
          input: ["text"],
          cost: { input: 0, output: 0 },
          contextWindow: 128000,
          maxTokens: 32000
        }]
      }
    }
  }
}

Important: baseUrl must include /v1 — OpenClaw automatically appends /responses.

Agent Configuration

Reference the provider using the providerId/modelId format. Set a generous timeoutSeconds since multi-agent pipelines take longer than a single LLM call:

{
  agents: {
    list: [{
      id: "kaiban-team",
      default: true,
      model: "kaiban-adapter/kaiban"
    }],
    defaults: {
      model: {
        primary: "kaiban-adapter/kaiban"
      },
      timeoutSeconds: 600
    }
  }
}