Retrieve AI Agent Log Node Details

Before You Begin

🔗 Use the GraphQL Playground to execute the queries in this guide.

Prerequisites

1. Authentication: Use a Service Account token (Personal Access Tokens are deprecated).
2. Permissions: You must be admin of the pipe/table.
3. Feature Toggle: The organization must have AI Agents feature enabled via toggle.
4. Log UUID: Identify the execution log UUID. Use the aiAgentLogsByRepo query to list logs.
5. Node ID: Identify the specific node ID within the log's tracing graph. We are temporarily using an index but will eventually migrate to use an uuid.

Step 1: Find Your Log UUID and Node ID

See our Get resource IDs page for how to find the pipe UUID.

Via GraphQL Query

• Use the aiAgentLogsByRepo query to list logs and find the log UUID:

{
  aiAgentLogsByRepo(repoUuid: "pipe-uuid-123") {
    edges {
      node {
        uuid
        agentName
        status
        createdAt
      }
    }
  }
}

• Use the aiAgentLogDetails query to list the tracing nodes and find the node ID:

{
  aiAgentLogDetails(uuid: "log-uuid-123") {
    uuid
    tracingNodes {
      nodeName
      status
      message
    }
  }
}

Step 2: Query AI Agent Log Node Details

Use the aiAgentLogNodeDetails query to retrieve detailed information about a specific node in an AI agent behavior execution log.

{
  aiAgentLogNodeDetails(logUuid: "log-uuid-123", nodeId: 1) {
    nodeName
    status
    message
    credits
    evidence
    planningDetails
  }
}

Arguments

Key Fields Explained

• nodeName: Name of the tracing node (e.g., the action or tool that ran).
• status: Outcome of the node execution. See Status Values below.
• message: Reasoning or error message from the node (populated from reasoning or error_message).
• credits: Credits consumed by the node execution (if applicable).
• evidence: JSON payload containing chunking data from the document processing nodes.
• planningDetails: JSON payload with the node's internal planning information.

Status Values

The status field can have the following values:

Step 3: Execute and Interpret the Response

After running the query, you'll receive a structured JSON response. Here's an example of a successfully executed node:

{
  "data": {
    "aiAgentLogNodeDetails": {
      "nodeName": "Execution Plan",
      "status": "success",
      "message": "Step completed successfully.",
      "credits": 1.5,
      "evidence": null,
      "planningDetails": {
        "{'actions':['Analyze the emotional tone of the comment 'Bad pizza sauce' and classify it as negative.','Update the card with a suggestion to address the negative comment about the pizza sauce.','Update the card with 'no' for the question 'Did you like our product?' since the sentiment is negative.'],'summary':'The task is to analyze the emotional tone of a customer comment about the service and update the system with the sentiment classification and a suggestion to address the comment.'}"
      }
    }
  }
}

This query provides granular details about an individual node in the AI agent execution graph, which is useful for debugging failures and understanding the agent's reasoning at each step.