README.md

---
license: apache-2.0
task_categories:
- text-generation
tags:
- web-agent
- evaluation
- servicenow
- workarena
pretty_name: ServiceNow Tasks
dataset_info:
  features:
  - name: benchmark
    dtype: string
  - name: task_name
    dtype: string
  - name: task_seed
    dtype: int64
  - name: task_config
    dtype: string
  - name: goal
    dtype: string
  - name: start_url
    dtype: string
  - name: validation
    dtype: string
  splits:
  - name: train
    num_bytes: 3336790
    num_examples: 330
  download_size: 164187
  dataset_size: 3336790
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
---

# ServiceNow Tasks

An evaluation dataset for web agents operating on ServiceNow instances. Each row contains a task goal, a configuration dict, and a standalone Python validation function that scores agent performance by querying the ServiceNow API.

Derived from the [WorkArena](https://github.com/ServiceNow/WorkArena) L1 benchmark.

## Dataset overview

- **330 samples** across **33 task types** grouped into **8 categories**
- Each task has 10 seeded variants
- Validators are self-contained Python functions that call the ServiceNow REST API to verify outcomes

| Category | Task types | Samples | Description |
|---|---|---|---|
| `all-menu` | 1 | 10 | Navigate to a specific module via the application menu |
| `create` | 5 | 50 | Create a record (incident, change request, hardware asset, problem, user) |
| `filter` | 6 | 60 | Apply filters on list views (assets, incidents, users, etc.) |
| `sort` | 6 | 60 | Sort list views by one or more columns |
| `order` | 9 | 90 | Order items from the hardware service catalog |
| `impersonation` | 1 | 10 | Impersonate a specific user |
| `knowledge` | 1 | 10 | Answer a question using the knowledge base |
| `dashboard` | 4 | 40 | Read values from charts and dashboards |

## Schema

| Column | Type | Description |
|---|---|---|
| `benchmark` | `string` | Benchmark name (`workarena_l1`) |
| `task_name` | `string` | Fully qualified task name (e.g. `workarena.servicenow.create-incident`) |
| `task_seed` | `int` | Random seed for the task variant |
| `task_config` | `string` | JSON-encoded dict with task-specific parameters (expected values, URLs, field names, etc.) |
| `goal` | `string` | Natural language instruction shown to the agent |
| `start_url` | `string` | URL where the agent should begin |
| `validation` | `string` | Python source code of the validator module |

## Validation

Each validator module exposes a `validate(messages, config)` function:

```python
def validate(messages: List[Dict[str, str]], config: Dict[str, Any]) -> int:
    """
    Args:
        messages: Conversation history (list of {"role": ..., "content": ...} dicts).
                  The last assistant message is used for answer-based tasks.
        config:   The task_config dict for this sample.

    Returns:
        1 if the task was completed successfully, 0 otherwise.

    Raises:
        ValueError: If the task configuration is invalid (not the agent's fault).
    """
```

Validators for `create`, `filter`, `sort`, `order`, and `dashboard` tasks call the ServiceNow REST API to verify outcomes. The following environment variables must be set:

| Variable | Description |
|---|---|
| `SERVICENOW_INSTANCE_URL` | Base URL of the ServiceNow instance (e.g. `https://myinstance.service-now.com`) |
| `SERVICENOW_INSTANCE_UNAME` | ServiceNow username |
| `SERVICENOW_INSTANCE_PWD` | ServiceNow password |
| `EXTRA_HTTP_HEADERS` | Optional JSON string of additional headers (e.g. `{"X-Custom": "value"}`) |

### Usage

```python
import ast
from datasets import load_dataset

ds = load_dataset("servicenow-ai/servicenow-tasks")

sample = ds["train"][0]
config = ast.literal_eval(sample["task_config"])

# Load the validator
exec(sample["validation"], globals())
score = validate(
    messages=[{"role": "assistant", "content": "The answer is 42."}],
    config=config,
)
```

## Task categories

### All menu
Navigate to a target module (e.g. *Performance Analytics > Indicator Groups*). Validated by checking the final URL against the expected path.

### Create
Create a new record on a specific table with prescribed field values. Validated by querying the ServiceNow Table API to check that the record exists with the correct field values.

### Filter
Apply AND/OR filters on a list view. Validated by querying the ServiceNow Table API with the expected filter conditions and comparing the result set with the currently displayed list.

### Sort
Sort a list view by one or more columns in ascending or descending order. Validated by querying the ServiceNow Table API with the expected sort order and comparing against the displayed list.

### Order
Order an item from the service catalog with a specific quantity and configuration. Validated by querying the ServiceNow Table API to check that the catalog request was submitted with the correct item, quantity, and configuration options.

### Impersonation
Impersonate a specific user. Validated by checking the ServiceNow session API for the impersonated user.

### Knowledge base search
Answer a question using ServiceNow's knowledge base. Validated by checking that the agent's answer contains the expected string (or one of the alternative answers).

### Dashboard / Chart retrieval
Read a value, percentage, min, max, mean, median, or mode from a chart on a dashboard or report. Validated by fetching chart data from the ServiceNow Stats API and comparing against the agent's answer. Supports single-series charts, multi-series charts with `chart_series` filtering, and trend charts with time-period bucketing.