RhesisClient

The RhesisClient is the central entry point for the Rhesis SDK. It initializes telemetry, manages the connector for remote testing, and provides access to the Rhesis API.

Initialization

app.py
from rhesis.sdk import RhesisClient

client = RhesisClient(
    api_key="your-api-key",
    project_id="your-project-id",
    environment="development",
)

Parameters

Parameter	Required	Description
`api_key`	No	API key for authentication. Falls back to `RHESIS_API_KEY` env var.
`base_url`	No	API base URL. Falls back to `RHESIS_BASE_URL` env var. Default: `https://api.rhesis.ai`
`project_id`	No	Project ID for endpoint registration and tracing. Falls back to `RHESIS_PROJECT_ID` env var, then to the project bound to the API token.
`environment`	No	Environment name. Falls back to `RHESIS_ENVIRONMENT` env var. Default: `development`

Project ID Resolution

The client resolves project_id in the following order:

Explicit parameter passed to the constructor
RHESIS_PROJECT_ID environment variable
Token introspection - if the API token is scoped to a project, the client automatically resolves the project ID from the token

If a project-scoped token is used and an explicit project_id is also provided, the values must match. A project-scoped token cannot be used to access a different project.

Organization-scoped tokens (without a project boundary) require an explicit project_id via the constructor or environment variable.

The RhesisClient must be initialized before using @observe or @endpoint decorators. Without it, a RuntimeError is raised.

The WebSocket connector is lazy: it is created only when @endpoint is used.

Environment Variables

Configure the client entirely through environment variables:

terminal
export RHESIS_API_KEY="your-api-key"
export RHESIS_PROJECT_ID="your-project-id"
export RHESIS_ENVIRONMENT="development"
export RHESIS_BASE_URL="https://api.rhesis.ai"  # optional

app.py
from rhesis.sdk import RhesisClient

# All configuration from environment variables
client = RhesisClient()

Environment Values

The environment parameter determines where your endpoints and traces are registered:

Value	Use Case
`development`	Local development and testing
`staging`	Pre-production validation
`production`	Live systems

Disabling the Client

To disable all SDK functionality (useful for CI/CD or testing):

terminal
export RHESIS_CONNECTOR_DISABLED=true

Accepted values: true, 1, yes, on (case-insensitive)

When disabled:

@endpoint and @observe decorators return functions unmodified
No telemetry initialization occurs
No WebSocket connection is established
All SDK method calls become no-ops

Usage Patterns

Tracing Only

For observability without remote testing:

app.py
from rhesis.sdk import RhesisClient, observe

client = RhesisClient(
    api_key="your-api-key",
    project_id="your-project-id",
    environment="production",
)

@observe.llm(provider="openai", model="gpt-4")
def generate(prompt: str) -> str:
    return openai.chat.completions.create(...)

Remote Testing

For registering functions as testable endpoints:

app.py
from rhesis.sdk import RhesisClient, endpoint

client = RhesisClient(
    api_key="your-api-key",
    project_id="your-project-id",
    environment="development",
)

@endpoint()
def chat(input: str, conversation_id: str = None) -> dict:
    # Automatically traced and registered as endpoint
    return {"output": process(input), "conversation_id": conversation_id}

Next steps

Register a function as a remote endpoint with the connector
Instrument LLM calls with the @observe decorator, covered in tracing
Work with tests and test sets through entities