Make AI work for Everyone. [![GenAI Engine CI](https://github.com/arthur-ai/arthur-engine/actions/workflows/arthur-engine-workflow.yml/badge.svg)](actions?query=workflow%3A%22%22GenAI+Engine+Workflow%22%22++) [![Discord](https://img.shields.io/badge/Discord-Arthur-blue?logo=discord&logoColor=white)](https://discord.gg/tdfUAtaVHz) [Website](https://arthur.ai?utm_source=github&utm_medium=readme) - [Documentation](https://docs.arthur.ai/?utm_source=github&utm_medium=readme) - [Talk to someone at Arthur](https://www.arthur.ai/book-demo?utm_source=github&utm_medium=readme)

# The Arthur Engine The Arthur Engine provides a **complete service** for developing, monitoring, and governing your AI/ML workloads using popular open-source technologies and frameworks. It is a tool designed for: - **Enforcing Guardrails in your LLM Applications and Generative AI Workflows** - Configurable metrics for real-time detection of PII or Sensitive Data leakage, Hallucination, Prompt Injection attempts, Toxic language, and other quality metrics - **Building, Evaluating, Observing and Governing Agentic Applications** - Collect and analyze OpenInference OpenTelemetry (OTEL) traces from any agentic workflow or LLM application - Run continuous evaluations on live traces to catch regressions and quality issues automatically - Manage, version, and iterate on prompts across your applications - Run experiments to compare prompt variants and measure their impact on quality metrics - Evaluate and monitor Retrieval-Augmented Generation (RAG) pipelines end-to-end - **Monitoring and Benchmarking Machine Learning models (requires the Arthur Platform)** - Support for a wide range of evaluation metrics (e.g., drift, accuracy, precision, recall, F1, and AUC) - Tools for comparing models, exploring feature importance, and identifying areas for optimization - For LLMs/GenAI applications, measure and monitor response relevance, hallucination rates, token counts, latency, and more - **Extensibility to fit into your application's architecture** - Native support for custom metrics and extensible API ## Quickstart **Claude Code users** Paste this prompt directly into Claude Code that's running on your agent application: ``` For each skill name in this list — arthur-onboard-oss, arthur-onboard-oss-engine, arthur-onboard-task, arthur-onboard-analyze, arthur-onboard-instrument, arthur-onboard-prompts, arthur-onboard-verify, arthur-onboard-eval-provider, arthur-onboard-evals, arthur-skills-upgrade — fetch https://raw.githubusercontent.com/arthur-ai/arthur-engine/refs/heads/main/integrations/claude-code-skills/arthur-onboard//SKILL.md and save it to ~/.claude/skills//SKILL.md (create the directory if it doesn't exist). Also fetch https://raw.githubusercontent.com/arthur-ai/arthur-engine/refs/heads/main/integrations/claude-code-skills/arthur-onboard/arthur-onboard-instrument/EXAMPLES.md and save it to ~/.claude/skills/arthur-onboard-instrument/EXAMPLES.md. Once all files are saved, read ~/.claude/skills/arthur-onboard-oss/SKILL.md and follow its instructions. ``` **Codex users** Paste this prompt directly into a Codex session that's running on your agent application: ``` Install the Arthur OSS onboarding Codex skills from arthur-ai/arthur-engine. Install these skill folders into ${CODEX_HOME:-~/.codex}/skills: - arthur-onboard-oss - arthur-onboard-oss-engine - arthur-onboard-task - arthur-onboard-analyze - arthur-onboard-instrument - arthur-onboard-prompts - arthur-onboard-verify - arthur-onboard-eval-provider - arthur-onboard-evals Fetch each SKILL.md from: https://raw.githubusercontent.com/arthur-ai/arthur-engine/main/integrations/claude-code-skills/arthur-onboard//SKILL.md Create each directory as needed. After installation, remind me to restart Codex. ``` **Everyone else** 1. Run the engine installer with the below command: Mac ``` bash <(curl -sSL https://get-genai-engine.arthur.ai/mac) ``` Windows ``` iex (iwr -Uri "https://get-genai-engine.arthur.ai/win" -UseBasicParsing).Content ``` 2. Instrument your agents for evaluations and LLM guardrailing by referencing the examples: https://github.com/arthur-ai/arthur-engine/tree/dev/genai-engine/examples ![Arthur Engine](./docs/images/arthur-engine.png) ## Arthur Platform Free Version To unlock the full capabilities of the Arthur Platform, [sign up](https://platform.arthur.ai/signup) and get started for free. * Custom dashboards * Alerts and notifications * Configurable webhook that can trigger any workflow * Agent discovery * Governance **Claude Code users** Paste this prompt directly into a Claude Code session that's running on your agent application: ``` For each skill name in this list — arthur-onboard-platform, arthur-onboard-platform-access, arthur-onboard-platform-workspace, arthur-onboard-platform-engine, arthur-onboard-platform-model, arthur-onboard-platform-token, arthur-onboard-analyze, arthur-onboard-instrument, arthur-onboard-prompts, arthur-onboard-verify, arthur-onboard-eval-provider, arthur-onboard-evals — fetch https://raw.githubusercontent.com/arthur-ai/arthur-engine/refs/heads/main/integrations/claude-code-skills/arthur-onboard//SKILL.md and save it to ~/.claude/skills//SKILL.md (create the directory if it doesn't exist). Once all skills are saved, read ~/.claude/skills/arthur-onboard-platform/SKILL.md and follow its instructions. ``` **Codex users** Install the Arthur Platform onboarding Codex skills from arthur-ai/arthur-engine. ``` Install these skill folders into ${CODEX_HOME:-~/.codex}/skills: - arthur-onboard-platform - arthur-onboard-platform-access - arthur-onboard-platform-workspace - arthur-onboard-platform-engine - arthur-onboard-platform-model - arthur-onboard-platform-token - arthur-onboard-analyze - arthur-onboard-instrument - arthur-onboard-prompts - arthur-onboard-verify - arthur-onboard-eval-provider - arthur-onboard-evals Fetch each SKILL.md from: https://raw.githubusercontent.com/arthur-ai/arthur-engine/main/integrations/claude-code-skills/arthur-onboard//SKILL.md Create each directory as needed. After installation, remind me to restart Codex. ``` **Everyone else** Onboard via the platform UI. ![Arthur Platform](./docs/images/arthur-platform.png) ## Arthur Platform Enterprise Version **The enterprise version of the Arthur Platform provides better performance, additional features, and capabilities**, including custom enterprise-ready guardrails + metrics, which can maximize the potential of AI for your organization. Key features: - State-of-the-art proprietary evaluation models trained by Arthur's world-class machine learning engineering team - Air-gapped deployment of the Arthur Engine (no dependency to Hugging Face Hub) - Optional on-premises deployment of the entire Arthur Platform - Support from the world-class engineering teams at Arthur To learn more about the enterprise version of the Arthur Platform, [reach out!](https://www.arthur.ai/book-demo) ## Contributing - Join the Arthur community on [Discord](https://discord.gg/tdfUAtaVHz) to get help and share your feedback. - To make a request for a bug fix or a new feature, please file a [GitHub issue](https://github.com/arthur-ai/arthur-engine/issues). - To make code contributions, please review the [contributing guidelines](CONTRIBUTING.md). - Thank you!