WhoAmI¶

The WhoAmI tool gives an agent self-awareness about its own identity within the platform. It returns the agent's workflow slug, node ID, current system prompt, extra configuration, and step-by-step instructions for self-modification.

Component type: whoami

How It Works¶

When invoked, the tool follows the tool edge from itself back to the parent agent node, then queries the database for the agent's current configuration and the workflow it belongs to. The response includes everything the agent needs to modify itself via the Platform API.

Ports¶

Outputs¶

Output Format¶

{
  "identity": {
    "workflow_slug": "my-workflow",
    "workflow_id": 42,
    "node_id": "agent_abc123",
    "component_type": "agent"
  },
  "current_config": {
    "system_prompt": "You are a helpful assistant...",
    "system_prompt_length": 45,
    "extra_config": {
      "conversation_memory": true
    }
  },
  "self_modification": {
    "endpoint": "/api/v1/workflows/my-workflow/nodes/agent_abc123/",
    "method": "PATCH",
    "example_body": {
      "config": {
        "system_prompt": "Your new system prompt here",
        "extra_config": { "conversation_memory": true }
      }
    },
    "instructions": [
      "1. Use create_agent_user to get API credentials if you don't have them",
      "2. Use platform_api with method='PATCH' to update your configuration",
      "3. Changes take effect on the next execution/conversation"
    ]
  }
}

Configuration¶

This tool has no configurable settings. It automatically resolves its parent agent from the tool edge connection.

Usage¶

Connect this tool to an agent via the green diamond tool handle. It is most useful alongside create_agent_user and platform_api:

flowchart LR
    WhoAmI[WhoAmI] -.->|tool| Agent
    CAU[Create Agent User] -.->|tool| Agent
    PA[Platform API] -.->|tool| Agent
    Model[AI Model] -.->|llm| Agent
    Trigger[Chat Trigger] --> Agent

Tool Signature¶

whoami() -> str

This tool takes no parameters. It automatically discovers the parent agent node through the edge graph.

Example¶

A self-improving agent's system prompt might include:

You are a self-improving assistant. When asked to change your behavior:
1. Call whoami() to learn your current configuration.
2. Call create_agent_user() to get API credentials.
3. Use platform_api() to PATCH your system prompt at the endpoint from whoami.
4. Confirm to the user that the change will take effect on the next conversation.

When a user says "Be more concise in your responses," the agent:

1. Calls whoami() to get its current system prompt and the PATCH endpoint.
2. Calls create_agent_user() to get an API key.
3. Calls platform_api(method="PATCH", path="/api/v1/workflows/my-workflow/nodes/agent_abc123/", body='{"config": {"system_prompt": "You are a concise assistant..."}}', api_key="...").