Submit
Icon for Archestra

Archestra

AGPL-3.0 hybrid platform combining LLM gateway, MCP orchestrator, and chat UI into one Kubernetes-native stack. Orchestrator spawns each MCP server as an isolated K8s pod — strong isolation, but K8s-only in practice.

Screenshot of Archestra website

Archestra is an AGPL-3.0 licensed enterprise MCP platform that provides a hybrid LLM gateway plus MCP orchestrator plus chat UI in one stack. Its distinguishing feature is that the orchestrator spawns each MCP server as an isolated Kubernetes pod, which provides strong process isolation but makes the platform effectively Kubernetes-dependent in practice.

Core capabilities

  • LLM proxy over mainstream providers with routing and cost tracking.
  • MCP orchestrator that spawns MCP servers as isolated Kubernetes pods with network policy enforcement.
  • Built-in chat UI for end users to interact with LLMs and their available MCP tools.
  • RBAC for users, teams, and MCP tool access.
  • Audit logging of all tool invocations with configurable retention.
  • SSO integration with OIDC providers.

License consideration

AGPL-3.0 is a strong copyleft license. Internal self-hosted use without redistribution does not trigger source-disclosure obligations, so private enterprise deployments are unencumbered. Exposing a modified version as a public or customer-facing network service triggers the AGPL network clause and requires source publication.

Deployment

Kubernetes-first. The orchestrator uses Kubernetes APIs to schedule MCP pods, which means Docker Swarm is not a supported target in practice — the "K8s-only" constraint is architectural, not packaging.

Limitations

  • Kubernetes dependency is architectural rather than optional — Docker Swarm and single-VM deployments are not supported.
  • AGPL-3.0 makes redistribution or SaaS rehosting legally complex.
  • Young startup project — review commit velocity, issue response, and sustainability before relying on it for production.
  • The hybrid LLM-plus-MCP-plus-chat scope is broad, which means each axis is less deep than a purpose-built tool (LiteLLM for LLM proxy, ContextForge for MCP, LibreChat for chat UI).
  • Smaller community and ecosystem than better-known gateways.

When to pick it

Kubernetes-native teams that want a single vendor to cover LLM proxy, MCP orchestration, and chat UI under one roof, and that are comfortable with AGPL-3.0 for internal use. Not a fit for Docker Swarm or Docker-only deployments.

Share:

Kind
Platform
Vendor
Archestra AI
License
Open Source
Website
archestra.ai
Deployment Type
Show all
Active
Ad
Icon

 

  
 

Similar to Archestra

Icon

 

  
  
Icon

 

  
  
Icon