Between October and early November 2025, the three official client libraries for the RAIL Score API went public. All three are MIT-licensed and maintained from the Responsible AI Labs GitHub organization. A dedicated Open Source hub collects install commands, quick-start snippets, and repository links.
Python SDK
v1.0.0 on Oct 18, 2025
pip install rail-score-sdkSync and async clients, compliance checks, safe regeneration, policy engine, middleware, agent evaluation, and full OpenTelemetry observability. Provider extras for OpenAI, Anthropic, Google Generative AI, and LiteLLM.
View on PyPI →JavaScript / TypeScript SDK
First release Nov 3, 2025
npm install @responsible-ai-labs/rail-scoreType-safe API coverage for eval, safe-regenerate, compliance, and health. Built-in wrappers for OpenAI, Anthropic, and Google. Policy engine, session tracking, Langfuse integration, and OTEL telemetry.
View on npm →Drupal module
Project registered Nov 4, 2025
composer require drupal/rail_scoreDrupal 9/10/11 integration: automatic RAIL evaluation on entity presave, compliance checks against six regulatory frameworks (GDPR, HIPAA, EU AI Act, CCPA, India DPDP, India AI Governance), structured audit logging, incident tracking, and a persistent human review queue. Optional integration with the Drupal AI module scores every AI-generated response automatically.
View on drupal.org →All three clients share the same versioning: SDK v2.4.0 as of March 2026 added agent evaluation, pre/post tool-call risk assessment, prompt injection detection, and stateful agent sessions. Release notes live in each package's repository.