Skip to main content

Documentation Index

Fetch the complete documentation index at: https://mintlify.com/archestra-ai/archestra/llms.txt

Use this file to discover all available pages before exploring further.

Archestra gives platform and development teams a single control plane for enterprise AI. Deploy the platform with one Docker command, then connect your LLM providers, install MCP servers from the private registry, build agents, and enforce security policies — all without writing custom infrastructure code.

Quickstart

Run Archestra locally in minutes with a single Docker command and build your first AI agent.

Deployment

Deploy to production using Helm on Kubernetes, with Terraform and Crossplane support.

MCP Registry & Gateway

Centralize MCP servers in a private registry and expose them through a governed gateway.

AI Agents

Build autonomous agents with no-code, assign MCP tools and knowledge bases, set triggers.

LLM Proxy

Drop-in proxy between AI apps and LLM providers with cost tracking and security policies.

Security & Guardrails

Deterministic tool call policies and Dual LLM isolation that survive prompt injection attacks.

Observability

Prometheus metrics, OpenTelemetry traces, and Grafana dashboards for every agent and team.

API Reference

REST API for managing agents, MCP registry entries, gateways, and platform resources.

​
What is Archestra?

Archestra is an open-source enterprise AI platform built for organizations where software engineers and non-technical teams both need to work with AI agents. It is composed of modular components you can adopt individually or as a full stack:
  • MCP Registry & Orchestrator — A private catalog of approved MCP servers that run as isolated Kubernetes pods. Share tools org-wide without distributing credentials to individual machines.
  • MCP Gateway — A single governed endpoint for Cursor, Claude Desktop, Open WebUI, and custom agents. Clients authenticate once; Archestra resolves the right upstream credential per tool call.
  • LLM Proxy — A drop-in OpenAI-compatible proxy that sits between your apps and providers like OpenAI, Anthropic, and Gemini. Captures costs, enforces limits, and applies security policies.
  • Agentic Chat — A ChatGPT-like interface for non-technical users, with access to agents via Slack, Microsoft Teams, or email.
  • Agent Runtime — No-code builder for autonomous agents. Assign MCP tools, knowledge bases, and sub-agents. Configure schedule, webhook, or email triggers.
  • Knowledge Base — Built-in RAG with chunking, embedding, hybrid search, and reranking — no external vector database required.
  • AI Tool Guardrails — Deterministic tool call and result policies evaluated at the LLM Proxy before requests reach the model. Cannot be bypassed by prompt injection.
Archestra is a CNCF Sandbox project and a Linux Foundation member. It is licensed under AGPL-3.0 and available on GitHub.

​
Get Started

1

Run Archestra locally

Pull the all-in-one Docker image and start the platform with a single command. The admin UI is available at http://localhost:3000 and the API at http://localhost:9000.
docker pull archestra/platform:latest
docker run -p 9000:9000 -p 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
2

Install an MCP server

Go to MCP Registry, search for an MCP server (e.g., microsoft__playwright-mcp), and install it. Archestra automatically runs it as an isolated pod in the embedded Kubernetes cluster.
3

Build an agent

Go to Agents, create a new agent with a system prompt, assign MCP tools, and optionally add a Knowledge Base. The agent is ready to use from Chat, Slack, Teams, or the API.
4

Connect LLM providers

Go to Settings → LLM API Keys and add your preferred provider — OpenAI, Anthropic, Gemini, or a self-hosted model via Ollama or vLLM.
Ready for production? See the Deployment guide for Helm installation, external PostgreSQL, and cloud-specific configuration.

Build docs developers (and LLMs) love

Get started for freeTalk to us