> ## Documentation Index
> Fetch the complete documentation index at: https://productlasso.com/docs/llms.txt
> Use this file to discover all available pages before exploring further.

# Get row traces

> Retrieve AI extraction and enhancement traces for a row.

Traces provide full transparency into how AI generated or enhanced each column value, including the prompts, model used, confidence scores, and source references.

## Path parameters

<ParamField path="table_id" type="string" required>
  The unique identifier of the table.
</ParamField>

<ParamField path="row_id" type="string" required>
  The unique identifier of the row.
</ParamField>

## Query parameters

<ParamField query="column_key" type="string">
  Filter traces to a specific column. Omit to get traces for all columns.
</ParamField>

## Response

<ResponseField name="data" type="array">
  <Expandable title="Trace object">
    <ResponseField name="column_key" type="string">The column this trace belongs to.</ResponseField>
    <ResponseField name="model" type="string | null">The AI model used.</ResponseField>
    <ResponseField name="confidence" type="number | null">Confidence score (0-1).</ResponseField>

    <ResponseField name="request" type="object">
      <Expandable title="Request details">
        <ResponseField name="system_prompt" type="string | null">The system prompt sent to the model.</ResponseField>
        <ResponseField name="user_content" type="string | null">The user content sent to the model.</ResponseField>
        <ResponseField name="tools" type="array">Tools available to the model.</ResponseField>
      </Expandable>
    </ResponseField>

    <ResponseField name="response" type="object">
      <Expandable title="Response details">
        <ResponseField name="text" type="string | null">The raw model response.</ResponseField>
        <ResponseField name="iterations" type="integer | null">Number of reasoning iterations.</ResponseField>
        <ResponseField name="tool_calls" type="array">Tools called by the model.</ResponseField>
        <ResponseField name="thought_summary" type="string | null">Summary of the model's reasoning.</ResponseField>
        <ResponseField name="sources" type="array">Web sources used for grounding.</ResponseField>
      </Expandable>
    </ResponseField>
  </Expandable>
</ResponseField>

<RequestExample>
  ```typescript TypeScript theme={null}
  const traces = await client.enhance.traces(
    "tbl_abc123",
    "row_xyz789",
    "description"
  );

  for (const trace of traces.data) {
    console.log(trace.column_key);
    console.log(trace.model);
    console.log(trace.confidence);
  }
  ```

  ```python Python theme={null}
  traces = client.enhance.traces(
      "tbl_abc123",
      "row_xyz789",
      column_key="description",
  )

  for trace in traces["data"]:
      print(trace["column_key"])
      print(trace["model"])
      print(trace["confidence"])
  ```

  ```bash cURL theme={null}
  curl -X GET "https://hub.banditshq.com/api/v1/tables/tbl_abc123/rows/row_xyz789/traces?column_key=description" \
    -H "Authorization: Bearer lasso_..."
  ```
</RequestExample>

<ResponseExample>
  ```json Response theme={null}
  {
    "data": [
      {
        "column_key": "description",
        "model": "gemini-pro",
        "confidence": 0.92,
        "request": {
          "system_prompt": "You are a product data expert...",
          "user_content": "Generate a description for: iPhone 15 Pro",
          "tools": []
        },
        "response": {
          "text": "The iPhone 15 Pro features...",
          "iterations": 1,
          "tool_calls": [],
          "thought_summary": "Used product name and specs to generate description",
          "sources": []
        }
      }
    ]
  }
  ```
</ResponseExample>