MUXI AI Application Server

What does this agent do, and when should you use it?

MUXI is a self-hosted AI application server made up of the Server, Runtime, and CLI, with agent formations as declarative deployment units. A formation can package agents, knowledge, memory, tools, workflows, triggers, policies, and operational behavior; the Overlord routes requests, coordinates agents, and runs workflows. Deployed formations can be reached through REST, SSE, MCP, and twelve official SDKs. The platform includes scoped, auditable memory, multi-tenant isolation, group-based RBAC, observability, and resilience features. It suits teams that want to operate complete agent systems on their own infrastructure; the README documents a self-hosted path rather than a ready-made hosted service.

Users fetch a formation with muxi pull, define systems and agents in .afs YAML files, and deploy with muxi deploy. The Server and Runtime execute the formation: the Overlord routes requests, matches SOPs, decomposes complex work, and coordinates agents. A formation can retrieve local or remote knowledge, call tools and MCP services, and run workflows and triggers. Memory is stored as layered, scoped events with provenance, rebuilds, and selective forgetting. Users can chat with a deployed formation through muxi chat, REST, SSE, MCP, or an SDK and receive streamed responses; a formation can also be exposed as an MCP server to other clients.

1. A SaaS platform team adding AI features can define agents in formations and serve them with user isolation and memory from one server stack.
2. An organization deploying internal AI tools can configure group inheritance and access rules for agents, MCP servers, and individual tools.
3. A team answering questions from product documentation can configure knowledge sources with hierarchical and vector retrieval.
4. A team that needs agents to act on schedules or external events can configure heartbeats, active hours, scheduled jobs, or webhooks and route the results.
5. Developers connecting agent systems to other clients can expose a formation through MCP or call it through one of the twelve SDKs.

How do you install or deploy this agent?

The README documents CLI installation for macOS, Linux, and Windows. After installation, users pull and deploy a formation. Its example formation selects an OpenAI model and reads OPENAI_API_KEY; the README does not provide a complete list of server system dependencies.

How do you use this agent?

Install the CLI, then pull and deploy the example formation:

brew install muxi-ai/tap/muxi
muxi pull @muxi/hello-muxi
muxi deploy
muxi chat hello-muxi

Linux installation:

curl -fsSL https://muxi.org/install | sudo bash

Windows installation:

irm https://muxi.org/install | iex

Formations define models, keys, and agents in YAML. The example uses OpenAI, so the deployment environment must provide OPENAI_API_KEY:

schema: "1.0.0"
id: "customer-support"
description: Helps customers with product support and refunds
llm:
  models:
    - text: "openai/gpt-4o"
  api_keys:
    openai: "${{ secrets.OPENAI_API_KEY }}"
agents:
  - sales-assistant

The Python SDK example connects to a local server with FormationClient and a client key:

from muxi import FormationClient

client = FormationClient(
    server_url="http://localhost:7890",
    formation_id="hello-muxi",
    client_key="<your-client-key>"
)

for chunk in client.chat_stream({ "message": "Hello!" }, user_id="user_123"):
    print(chunk.get("text", ""), end="")

What are this agent's strengths and limitations?

How does this agent compare with similar options?

Key facts side by side with the most closely related agents.

How does FollowAgents rate this agent?

FAQ