pycredoai 1.1.0

Credo AI SDK

CredoAI Python SDK

Credo AI SDK

Quick Start

Using Environment Variables (Recommended)

Set the following environment variables:

export CREDOAI_API_KEY="your-api-key"
export CREDOAI_API_URL="https://api.credo.ai"  # optional, this is the default
export CREDOAI_TENANT="your-tenant"            # required, your CredoAI tenant

Then use the client without any arguments:

from credoai import CredoAI, UseCaseCreate

# No arguments needed when env vars are set
client = CredoAI()

# Create a use case
use_case = client.use_cases.create(
    data=UseCaseCreate(name="My AI Model", description="A machine learning model")
)
print(f"Created: {use_case.id}")

Synchronous Client

from credoai import CredoAI, UseCaseCreate, UseCaseUpdate

# Option 1: Using environment variables (recommended)
client = CredoAI()

# Option 2: Using explicit arguments (overrides env vars)
client = CredoAI(
    base_url="https://api.credo.ai",  # optional - defaults to CREDOAI_API_URL or https://api.credo.ai
    api_key="your-api-key",      # required if CREDOAI_API_KEY not set
    tenant="your-tenant"              # optional - defaults to CREDOAI_TENANT or "credoai"
)

# Create a use case with Pydantic model
use_case_data = UseCaseCreate(
    name="My AI Model",
    description="A machine learning model for predictions"
)
use_case = client.use_cases.create(data=use_case_data)

# Get a use case
use_case = client.use_cases.get(use_case_id="123")

# Update a use case with Pydantic model
update_data = UseCaseUpdate(description="Updated description")
updated_use_case = client.use_cases.update(use_case_id="123", data=update_data)

# Health checks
status = client.system.ping()
liveness = client.health.check_liveness()
readiness = client.health.check_readiness()

Asynchronous Client

import asyncio
from credoai import AsyncCredoAI, UseCaseCreate, UseCaseUpdate

async def main():
    # Option 1: Using environment variables (recommended)
    client = AsyncCredoAI()

    # Option 2: Using explicit arguments
    client = AsyncCredoAI(
        base_url="https://api.credo.ai",
        api_key="your-api-key"
    )

    # Create a use case with Pydantic model
    use_case_data = UseCaseCreate(
        name="My AI Model",
        description="A machine learning model for predictions"
    )
    use_case = await client.use_cases.create(data=use_case_data)

    # Get a use case
    use_case = await client.use_cases.get(use_case_id="123")

    # Update a use case with Pydantic model
    update_data = UseCaseUpdate(description="Updated description")
    updated_use_case = await client.use_cases.update(use_case_id="123", data=update_data)

    # Health checks
    status = await client.system.ping()
    liveness = await client.health.check_liveness()
    readiness = await client.health.check_readiness()

# Run async code
asyncio.run(main())

Configuration

Environment Variables

The SDK supports configuration via environment variables. This is the recommended approach for production deployments:

Variable	Description	Default
CREDOAI_API_KEY	API key for authentication	(required)
CREDOAI_API_URL	Base URL of the API service	https://api.credo.ai
CREDOAI_TENANT	Your CredoAI tenant identifier	(required)

Note: The SDK automatically appends /api/v1/integration to the base URL. You only need to provide the base domain (e.g., https://api.credo.ai).

# Set environment variables
export CREDOAI_API_KEY="your-api-key"
export CREDOAI_API_URL="https://api.credo.ai"
export CREDOAI_TENANT="your-tenant"

from credoai import CredoAI

# Client automatically uses environment variables
client = CredoAI()

Constructor arguments always take precedence over environment variables:

# Explicit arguments override env vars
client = CredoAI(
    api_key="your-api-key",  # Overrides CREDOAI_API_KEY
    tenant="other-tenant"    # Overrides CREDOAI_TENANT
    # base_url uses CREDOAI_API_URL since not specified
)

Authentication

The SDK uses API key authentication with automatic JWT token exchange:

from credoai import CredoAI

# Using environment variables (recommended)
client = CredoAI()

# Or with explicit arguments
client = CredoAI(
    base_url="https://api.credo.ai",
    api_key="your-api-key",
    tenant="your-tenant"
)

Authentication Flow

• The client calls /api/v1/integration/auth/token at initialization to exchange your API key for a JWT token
• All API requests use Authorization: Bearer <jwt-token> header format internally

Advanced Configuration

import httpx
from credoai import CredoAI

# Full configuration with all options
client = CredoAI(
    base_url="https://api.credo.ai",  # Or set CREDOAI_API_URL
    api_key="your-api-key",           # Or set CREDOAI_API_KEY
    tenant="your-tenant",             # Or set CREDOAI_TENANT
    timeout=30.0,
    verify_ssl=True,
    headers={"Custom-Header": "value"},
    cookies={"session": "token"}
)

# Custom httpx configuration
client = CredoAI(
    api_key="your-api-key",  # Other params from env vars
    httpx_args={
        "proxies": "http://localhost:8080",
        "event_hooks": {
            "request": [lambda request: print(f"Request: {request.method} {request.url}")],
            "response": [lambda response: print(f"Response: {response.status_code}")]
        }
    }
)

Context Managers

Both clients support context managers for automatic resource cleanup:

Synchronous

from credoai import CredoAI

with CredoAI(base_url="https://api.example.com", api_key="your-api-key") as client:
    use_case = client.use_cases.get(use_case_id="123")
    # Connection automatically closed

Asynchronous

from credoai import AsyncCredoAI

async def main():
    async with AsyncCredoAI(base_url="https://api.example.com", api_key="your-api-key") as client:
        use_case = await client.use_cases.get(use_case_id="123")
        # Connection automatically closed

asyncio.run(main())

API Examples

Use Cases

Create, list, read, update, and delete use cases.

Create a Use Case

from credoai import CredoAI, UseCaseCreate

client = CredoAI(base_url="https://api.example.com", api_key="your-api-key")

# Create with Pydantic model
data = UseCaseCreate(
    name="Risk Assessment Model",
    description="ML model for credit risk assessment"
)
use_case = client.use_cases.create(data=data)
print(f"Created use case with ID: {use_case.id}")

List Use Cases

from credoai import CredoAI

client = CredoAI(base_url="https://api.example.com", api_key="your-api-key")

# List all use cases
use_cases = client.use_cases.lists()
for uc in use_cases:
    print(f"- {uc.name}: {uc.description}")

Get a Use Case

from credoai import CredoAI

client = CredoAI(base_url="https://api.example.com", api_key="your-api-key")

# Get by ID
use_case = client.use_cases.get(use_case_id="123")
print(f"Name: {use_case.name}")
print(f"Description: {use_case.description}")

Update a Use Case

from credoai import CredoAI, UseCaseUpdate

client = CredoAI(base_url="https://api.example.com", api_key="your-api-key")

# Update with Pydantic model
data = UseCaseUpdate(
    name="Updated Risk Model",
    description="Updated description for the model"
)
updated = client.use_cases.update(use_case_id="123", data=data)
print(f"Updated: {updated.name}")

Delete a Use Case

from credoai import CredoAI

client = CredoAI(base_url="https://api.example.com", api_key="your-api-key")

# Delete by ID
client.use_cases.delete(use_case_id="123")
print("Use case deleted")

Models

Create, list, read, update, and delete models.

Create a Model

from credoai import CredoAI, ModelCreate

client = CredoAI(base_url="https://api.example.com", api_key="your-api-key")

# Create with Pydantic model
data = ModelCreate(
    name="GPT-4 Classification",
    description="LLM-based text classification model"
)
model = client.models.create(data=data)
print(f"Created model with ID: {model.id}")

List Models

from credoai import CredoAI

client = CredoAI(base_url="https://api.example.com", api_key="your-api-key")

# List all models
models = client.models.lists()
for m in models:
    print(f"- {m.name}: {m.description}")

Get a Model

from credoai import CredoAI

client = CredoAI(base_url="https://api.example.com", api_key="your-api-key")

# Get by ID
model = client.models.get(model_id="456")
print(f"Name: {model.name}")
print(f"Description: {model.description}")

Update a Model

from credoai import CredoAI, ModelUpdate

client = CredoAI(base_url="https://api.example.com", api_key="your-api-key")

# Update with Pydantic model
data = ModelUpdate(
    name="GPT-4 Classification v2",
    description="Improved classification model"
)
updated = client.models.update(model_id="456", data=data)
print(f"Updated: {updated.name}")

Delete a Model

from credoai import CredoAI

client = CredoAI(base_url="https://api.example.com", api_key="your-api-key")

# Delete by ID
client.models.delete(model_id="456")
print("Model deleted")

Vendors

Full CRUD operations for managing vendors.

Create a Vendor

from credoai import CredoAI, VendorCreate

client = CredoAI(base_url="https://api.example.com", api_key="your-api-key")

# Create with Pydantic model
data = VendorCreate(
    name="OpenAI",
    description="AI research and deployment company"
)
vendor = client.vendors.create(data=data)
print(f"Created vendor with ID: {vendor.id}")

List Vendors

from credoai import CredoAI

client = CredoAI(base_url="https://api.example.com", api_key="your-api-key")

# List all vendors
vendors = client.vendors.lists()
for v in vendors:
    print(f"- {v.name}: {v.description}")

Get a Vendor

from credoai import CredoAI

client = CredoAI(base_url="https://api.example.com", api_key="your-api-key")

# Get by ID
vendor = client.vendors.get(vendor_id="789")
print(f"Name: {vendor.name}")
print(f"Description: {vendor.description}")

Update a Vendor

from credoai import CredoAI, VendorUpdate

client = CredoAI(base_url="https://api.example.com", api_key="your-api-key")

# Update with Pydantic model
data = VendorUpdate(
    name="OpenAI Inc.",
    description="Updated vendor description"
)
updated = client.vendors.update(vendor_id="789", data=data)
print(f"Updated: {updated.name}")

Delete a Vendor

from credoai import CredoAI

client = CredoAI(base_url="https://api.example.com", api_key="your-api-key")

# Delete by ID
client.vendors.delete(vendor_id="789")
print("Vendor deleted")

System & Health

System status and health check endpoints.

Ping

from credoai import CredoAI

client = CredoAI(base_url="https://api.example.com", api_key="your-api-key")

# Check system status
status = client.system.ping()
print(f"System status: {status}")

Metrics

from credoai import CredoAI

client = CredoAI(base_url="https://api.example.com", api_key="your-api-key")

# Get system metrics
metrics = client.system.metrics_metrics()
print(f"Metrics: {metrics}")

Liveness Check

from credoai import CredoAI

client = CredoAI(base_url="https://api.example.com", api_key="your-api-key")

# Check if service is alive
liveness = client.health.check_liveness()
print(f"Liveness: {liveness}")

Readiness Check

from credoai import CredoAI

client = CredoAI(base_url="https://api.example.com", api_key="your-api-key")

# Check if service is ready to accept requests
readiness = client.health.check_readiness()
print(f"Readiness: {readiness}")

Async Examples

All endpoints are available with async versions using AsyncCredoAI:

import asyncio
from credoai import (
    AsyncCredoAI,
    UseCaseCreate, UseCaseUpdate,
    ModelCreate, ModelUpdate,
    VendorCreate, VendorUpdate,
)

async def main():
    client = AsyncCredoAI(base_url="https://api.example.com", api_key="your-api-key")

    # Use Cases
    use_case = await client.use_cases.create(
        data=UseCaseCreate(name="Async Use Case", description="Created async")
    )
    use_cases = await client.use_cases.lists()
    use_case = await client.use_cases.get(use_case_id=use_case.id)
    use_case = await client.use_cases.update(
        use_case_id=use_case.id,
        data=UseCaseUpdate(description="Updated async")
    )
    await client.use_cases.delete(use_case_id=use_case.id)

    # Models
    model = await client.models.create(
        data=ModelCreate(name="Async Model", description="Created async")
    )
    models = await client.models.lists()
    model = await client.models.get(model_id=model.id)
    model = await client.models.update(
        model_id=model.id,
        data=ModelUpdate(description="Updated async")
    )
    await client.models.delete(model_id=model.id)

    # Vendors
    vendor = await client.vendors.create(
        data=VendorCreate(name="Async Vendor", description="Created async")
    )
    vendors = await client.vendors.lists()
    vendor = await client.vendors.get(vendor_id=vendor.id)
    vendor = await client.vendors.update(
        vendor_id=vendor.id,
        data=VendorUpdate(description="Updated async")
    )
    await client.vendors.delete(vendor_id=vendor.id)

    # System & Health
    status = await client.system.ping()
    metrics = await client.system.metrics_metrics()
    liveness = await client.health.check_liveness()
    readiness = await client.health.check_readiness()

asyncio.run(main())

API Reference

Use Cases

Method	Description
use_cases.create(data: UseCaseCreate) -> UseCaseResponse	Create a new use case
use_cases.lists() -> List[UseCaseResponse]	List all use cases
use_cases.get(use_case_id: str) -> UseCaseResponse	Get a specific use case
use_cases.update(use_case_id: str, data: UseCaseUpdate) -> UseCaseResponse	Update a use case
use_cases.delete(use_case_id: str)	Delete a use case

Models

Method	Description
models.create(data: ModelCreate) -> ModelResponse	Create a new model
models.lists() -> List[ModelResponse]	List all models
models.get(model_id: str) -> ModelResponse	Get a specific model
models.update(model_id: str, data: ModelUpdate) -> ModelResponse	Update a model
models.delete(model_id: str)	Delete a model

Vendors

Method	Description
vendors.create(data: VendorCreate) -> VendorResponse	Create a new vendor
vendors.lists() -> List[VendorResponse]	List all vendors
vendors.get(vendor_id: str) -> VendorResponse	Get a specific vendor
vendors.update(vendor_id: str, data: VendorUpdate) -> VendorResponse	Update a vendor
vendors.delete(vendor_id: str)	Delete a vendor

System

Method	Description
system.ping() -> PingResponse	Health check / ping endpoint
system.metrics_metrics() -> Any	Get system metrics

Health

Method	Description
health.check_liveness() -> HealthResponse	Kubernetes liveness probe
health.check_readiness() -> HealthResponse	Kubernetes readiness probe

Client Configuration

Both CredoAI and AsyncCredoAI inherit from the base client classes and support these parameters:

Parameter	Type	Description	Default
base_url	str | None	Base URL for the API (without /api/v1/integration path)	CREDOAI_API_URL env var or https://api.credo.ai
api_key	str | None	Your API key for authentication	CREDOAI_API_KEY env var (required)
tenant	str | None	Your CredoAI tenant identifier	CREDOAI_TENANT env var (required)
timeout	float | httpx.Timeout	Request timeout configuration	30.0
verify_ssl	bool | str | ssl.SSLContext	SSL certificate verification	True
headers	dict[str, str]	Additional headers for all requests	{}
cookies	dict[str, str]	Cookies to include with all requests	{}
follow_redirects	bool	Whether to follow HTTP redirects	False
httpx_args	dict	Additional arguments passed to httpx client	{}

Low-Level API Access

For advanced users who need direct access to the underlying HTTP client:

from credoai import AuthenticatedClient
from credoai.models import UseCaseCreate
import httpx

# Using the lower-level authenticated client directly
client = AuthenticatedClient(
    base_url="https://api.example.com",
    token="your-api-key",
    auth_header_name="X-API-Key",
    prefix=""
)

# Manual HTTP request construction
use_case_data = UseCaseCreate(name="Test", description="A test use case")
json_data = use_case_data.model_dump()

with client as httpx_client:
    # Direct HTTP requests
    response = httpx_client.post("/api/v1/use_cases", json=json_data)
    print(f"Status: {response.status_code}, Data: {response.json()}")

    # Get request
    response = httpx_client.get("/api/v1/integration/ping")
    print(f"Ping: {response.json()}")

Error Handling

from credoai import CredoAI
from credoai.errors import ApiError
import httpx

client = CredoAI(base_url="https://api.example.com", api_key="your-api-key")

try:
    use_case = client.use_cases.get(use_case_id="invalid-id")
except ApiError as e:
    print(f"API error: {e.status_code} - {e.message}")
except httpx.HTTPStatusError as e:
    if e.response.status_code == 404:
        print("Use case not found")
    elif e.response.status_code == 422:
        print("Validation error")
        # Parse validation errors if needed
        if e.response.headers.get("content-type") == "application/json":
            error_data = e.response.json()
            print(f"Validation details: {error_data}")
    else:
        print(f"HTTP error: {e.response.status_code}")
except httpx.RequestError as e:
    print(f"Request failed: {e}")

Development

This package is generated using the Credo AI SDK generator.

To regenerate the client:

1. Ensure the API server is running with updated OpenAPI spec
2. Run the generation script from the parent repository
3. The client will be updated with any API changes

Contributing

1. Make changes to the API server and update OpenAPI specifications
2. Regenerate the client using the provided generation scripts
3. Test the generated client thoroughly
4. Submit pull requests to the main repository

License

This project is licensed under the terms specified in the parent repository.