Orchestrate Enterprise MCP Servers

Archestra is an all-in-one open-source enterprise AI platform built on a security and observability foundation. It bundles a chat interface for non-technical users (with projects, MCP apps, and Slack/MS Teams/email front-ends); a developer LLM and MCP portal giving one token for Claude Code, Codex, and Cursor; an LLM gateway supporting multiple providers (Anthropic, OpenAI, Azure, Bedrock, DeepSeek, and others) with cost limits, virtual API keys, and dynamic model routing; an MCP gateway with OAuth and On-Behalf-Of auth so tools run as the user rather than a shared service account; an A2A (agent-to-agent) gateway for webhook triggers; a private MCP registry for teams to ship their own tools; an MCP orchestrator with a Kubernetes operator and self-serve environment promotion; an agent runtime with scheduled/email/webhook triggers, sub-agent delegation, reusable skills, and sandboxed code execution; a RAG knowledge base via connectors; and a mini app builder.

Security and identity are first-class: deterministic tool-call guardrails, Dual-LLM verification, Lethal Trifecta protections, SSO (OIDC, SAML, Okta, Entra), RBAC with role mapping and team sync, secrets management, and per-environment egress and cost limits. Observability includes OpenTelemetry traces, Prometheus metrics, logs, and per-team cost tracking.

docker pull archestra/platform:latest

docker run \
  -p 127.0.0.1:9000:9000 -p 127.0.0.1:3000:3000 \
  -e ARCHESTRA_QUICKSTART=true \
  -v /var/run/docker.sock:/var/run/docker.sock \
  -v archestra-postgres-data:/var/lib/postgresql/data \
  -v archestra-app-data:/app/data \
  archestra/platform

Use this when an organization needs to centralize AI access behind one governed entry point: giving developers a single token across Claude Code, Codex, and Cursor, giving non-technical staff an internal chat assistant, and enforcing SSO/RBAC, cost limits, and guardrails across all of it. It targets enterprise deployment - it claims three Fortune-50 deployments, 31ms p95 latency, and includes a Terraform provider and Helm chart for production rollout.

Do NOT use this if you just need a single MCP tool connector for one assistant - Archestra is a full platform (gateway, orchestrator, agent runtime, identity layer) with real infrastructure to run (Docker/Kubernetes, Postgres), not a lightweight drop-in server. It also uses a dual licensing model (AGPL 3.0 / Enterprise), so confirm licensing terms fit your use case before adopting it.

• Chat interface: internal AI assistant with projects, MCP apps, Slack, MS Teams, and email front-ends
• LLM gateway: supports Anthropic, OpenAI, Azure, Bedrock, DeepSeek and others, with cost limits, virtual API keys, and dynamic model routing
• MCP gateway: OAuth + On-Behalf-Of auth so tools execute as the calling user
• A2A gateway: agent-to-agent webhook triggers
• Private MCP registry: teams publish their own internal MCP tools
• MCP orchestrator: Kubernetes operator with self-serve environment promotion
• Agent runtime: scheduled, email, and webhook triggers; sub-agent delegation; reusable skills; sandboxed code execution; Kubernetes-native filesystem
• RAG knowledge base: connectors into existing data stacks
• Mini app builder
• Guardrails: deterministic tool-call guardrails, Dual-LLM verification, Lethal Trifecta protections
• Identity & access: SSO (OIDC, SAML, Okta, Entra), RBAC with role mapping and team sync, secrets management
• Environments: per-environment egress policies and cost limits
• Observability: OpenTelemetry traces, Prometheus metrics, logs, per-team cost tracking

Quickstart via Docker:

docker pull archestra/platform:latest

docker run \
  -p 127.0.0.1:9000:9000 -p 127.0.0.1:3000:3000 \
  -e ARCHESTRA_QUICKSTART=true \
  -v /var/run/docker.sock:/var/run/docker.sock \
  -v archestra-postgres-data:/var/lib/postgresql/data \
  -v archestra-app-data:/app/data \
  archestra/platform

Then open http://localhost:3000. Production deployments have Docker, Helm chart, and Kubernetes instructions, plus a dedicated Terraform provider.

Platform, security, and infrastructure teams at organizations that need to govern AI tool and LLM access enterprise-wide - unifying developer coding assistants, internal chat, agent automation, and RAG under one identity-and-cost-controlled gateway.