niklub

"""
Simple demonstration of Data Processing Agent using MCP (Model Context Protocol) and BigQuery.
This is a tutorial example - not for production use.

To install, use [uv](https://github.com/astral-sh/uv):
```
uv add mcp[cli] google-cloud-bigquery pandas db-dtypes python-dotenv gspread
```

Environment setup in `.env.` file:

niklub

Random tasks assignments

Split tasks by users

class AssignmentSamplingMethod(Enum):
  RANDOM = auto() # produces uniform splits across annotators
  
project.assign_annotators_by_sampling(
  users: List[int],

niklub

{
  "id": 84720,
  "data": {
    "url": "https://htx-misc.s3.amazonaws.com/libri/dev-clean.2035.147960.2035-147960-0016.flac.wav"
  },
  "annotations": [{
    "id": 4200,
    "completed_by": {
      "id": 26,
      "email": "amy@macdonalds.com",

niklub

Project configuration

Go to “Create project”, and then paste the following config when prompted:

<View>
  <Header value="Read the dialogue:"></Header>
  <Dialog name="test" value="$dialog"></Dialog>
  <Header value="Selected message is:"></Header>
    <Choices name="label" toName="test" showInline="true">