archestra-ai/archestra

What it solves

Archestra provides a comprehensive, secure, and observable platform for enterprises to deploy and manage AI agents and LLM workflows. It eliminates the need to build fragmented infrastructure for security, cost tracking, and tool integration when scaling AI across an organization.

How it works

The platform acts as a centralized gateway and runtime environment that integrates several core components:

• LLM & MCP Gateways: Manages connections to various providers (Anthropic, OpenAI, etc.) with dynamic routing, virtual API keys, and cost limits.
• Agent Runtime: Executes agents with support for sub-agent delegation, scheduled triggers, and sandboxed code execution.
• MCP Orchestrator: Uses a Kubernetes operator to manage Model Context Protocol (MCP) apps and a private registry for team-specific tools.
• Knowledge Base: Implements RAG via connectors to existing data stacks.
• Security Layer: Integrates SSO (OIDC, SAML), RBAC, and deterministic guardrails (like Dual-LLM verification) to ensure safe tool calls.
• Observability: Provides built-in OpenTelemetry traces, Prometheus metrics, and per-team cost tracking.

Who it’s for

Enterprise organizations and development teams that need to deploy AI agents and LLM applications with production-grade security, governance, and infrastructure management.

Highlights

• Enterprise-grade security: Built-in SSO, RBAC, and sandboxed code execution.
• MCP Integration: Full support for Model Context Protocol (MCP) with OAuth and On-Behalf-Of authentication.
• Multi-provider support: A single gateway for multiple LLM providers with cost limits and routing.
• K8s-native: Includes a Kubernetes operator and a K8s-native filesystem for agents.
• Observability: Native OpenTelemetry and Prometheus integration for deep system visibility.