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

# Create User Template

> Save a new eval agent template, from scratch or by snapshotting an existing eval agent.

### Headers

<ParamField header="authorization" type="string" required>
  Your API key for authentication.
</ParamField>

### Body Parameters

<ParamField body="name" type="string">
  Display name for the template. Maximum 120 characters.
</ParamField>

<ParamField body="description" type="string">
  Description of the template. Maximum 4000 characters. Pass `null` to leave blank.
</ParamField>

<ParamField body="category" type="string">
  Category label for the template. Maximum 64 characters. Pass `null` to leave blank.
</ParamField>

<ParamField body="modality" type="string">
  Evaluation modality. Either `text` or `audio`.
</ParamField>

<ParamField body="system_prompt_md" type="string">
  System prompt for the eval agent, in Markdown. Maximum 50000 characters.
</ParamField>

<ParamField body="prompt_md" type="string">
  Evaluation prompt, in Markdown. Maximum 50000 characters.
</ParamField>

<ParamField body="levels" type="array">
  Ordered scoring levels. Each level object requires `level_key` (string, 1-64 chars), `label` (string, 1-80 chars), and `prompt_md` (string). The `color` field is optional: one of `rose`, `amber`, `gold`, `emerald`, `blue`, `indigo`, `violet`, or `fog`.
</ParamField>

<ParamField body="target_level_keys" type="array">
  Array of `level_key` strings that represent the passing threshold.
</ParamField>

<ParamField body="visibility" type="string">
  Access scope for the template. One of `private`, `org`, or `public`.
</ParamField>

<ParamField body="from_agent_id" type="string">
  UUID of an existing eval agent to snapshot content from.
</ParamField>

<ParamField body="from_version_id" type="string">
  UUID of a specific version of the eval agent to snapshot. Only used when `from_agent_id` is provided. Defaults to the agent's current version if omitted.
</ParamField>

<Note>
  There are two creation paths. Pass `from_agent_id` (optionally with `from_version_id`) to snapshot an existing agent's content into the template, or pass inline content fields to build one from scratch. When both are supplied, `from_agent_id` provides the content and inline fields act as overrides.
</Note>

### Response

<ResponseField name="id" type="string">
  Unique identifier (UUID) for the new template.
</ResponseField>

<ResponseField name="org_id" type="string">
  UUID of the organization that owns this template.
</ResponseField>

<ResponseField name="key" type="string">
  Short key used to reference the template.
</ResponseField>

<ResponseField name="name" type="string">
  Display name of the template.
</ResponseField>

<ResponseField name="description" type="string">
  Description of the template, or `null`.
</ResponseField>

<ResponseField name="category" type="string">
  Category of the template, or `null`.
</ResponseField>

<ResponseField name="modality" type="string">
  Evaluation modality. Either `text` or `audio`.
</ResponseField>

<ResponseField name="system_prompt_md" type="string">
  System prompt for the eval agent, in Markdown.
</ResponseField>

<ResponseField name="prompt_md" type="string">
  Evaluation prompt, in Markdown.
</ResponseField>

<ResponseField name="levels" type="array">
  Ordered scoring levels.
</ResponseField>

<ResponseField name="target_level_keys" type="array">
  Array of `level_key` strings that represent the passing threshold.
</ResponseField>

<ResponseField name="visibility" type="string">
  Access scope. One of `private`, `org`, or `public`.
</ResponseField>

<ResponseField name="source_agent_id" type="string">
  UUID of the eval agent this template was snapshotted from, or `null`.
</ResponseField>

<ResponseField name="source_version_id" type="string">
  UUID of the specific version snapshotted, or `null`.
</ResponseField>

<ResponseField name="created_by" type="string">
  UUID of the user who created the template, or `null`.
</ResponseField>

<ResponseField name="created_at" type="string">
  ISO 8601 timestamp of when the template was created.
</ResponseField>

<ResponseField name="updated_at" type="string">
  ISO 8601 timestamp of when the template was last updated.
</ResponseField>

<ResponseExample>
  ```json Response theme={null}
  {
    "data": {
      "id": "a1b2c3d4-5678-9abc-def0-1234567890ab",
      "org_id": "b2c3d4e5-6789-abcd-ef01-234567890abc",
      "key": "my_hallucination_check",
      "name": "My Hallucination Check",
      "description": "Customized hallucination detection for our product domain.",
      "category": "quality",
      "modality": "text",
      "system_prompt_md": "You are an expert evaluator assessing whether an AI agent fabricated information.",
      "prompt_md": "Review the conversation and identify any claims that were factually incorrect or unsupported.",
      "levels": [
        {
          "level_key": "no_hallucination",
          "label": "No Hallucination",
          "prompt_md": "The agent made no factually incorrect or unsupported claims.",
          "color": "emerald"
        },
        {
          "level_key": "hallucination_detected",
          "label": "Hallucination Detected",
          "prompt_md": "The agent stated something factually incorrect or unsupported.",
          "color": "rose"
        }
      ],
      "target_level_keys": ["no_hallucination"],
      "visibility": "org",
      "source_agent_id": null,
      "source_version_id": null,
      "created_by": "9a8b7c6d-5e4f-3210-9876-543210fedcba",
      "created_at": "2026-05-27T10:00:00.000Z",
      "updated_at": "2026-05-27T10:00:00.000Z"
    },
    "errors": null
  }
  ```
</ResponseExample>

***

Docs for agents: [llms.txt](/llms.txt)