Skip to content

Python SDK Quickstart

Run assessments programmatically using the Opencomplai Python SDK.

Install

Bash
pip install opencomplai
PowerShell
pip install opencomplai

Installing from source

The packages are published on PyPI. To work from a checkout instead, see Installation.

Minimal example

Python
from opencomplai import assess, AssessmentInput, ModelMetadata

assessment = AssessmentInput(
    model=ModelMetadata(
        name="my-model",
        version="1.0.0",
        modality="text",
        use_case="customer support chatbot",
        deployment_context="production",
    )
)

result = assess(assessment)

print(result.risk_level)          # "minimal"
print(result.evidence_summary)    # "Risk classification: MINIMAL. ..."
for rule in result.rule_results:
    print(rule.rule_id, rule.passed, rule.rationale)

Full output example

Python
from opencomplai import assess, AssessmentInput, ModelMetadata

result = assess(AssessmentInput(
    model=ModelMetadata(
        name="loan-decision-model",
        version="2.3.1",
        modality="text",
        use_case="automated credit scoring for retail lending",
        deployment_context="production",
    )
))

print(f"Risk level: {result.risk_level}")
print(f"Rules evaluated: {result.rules_evaluated}")
print(f"Rules passed: {result.rules_passed}")
print(f"Rules failed: {result.rules_failed}")
print(f"Evidence: {result.evidence_summary}")
print()
for r in result.rule_results:
    status = "" if r.passed else ""
    print(f"{status} [{r.rule_id}] {r.rule_name}")
    print(f"   Rationale: {r.rationale}")
    print(f"   Reference: {r.reference}")

API Reference

See SDK API Reference for full type signatures and model documentation.