SkillFoxx

MUXI: skill for the AI agent platform

MUXI

A skill for the MUXI server: the agent guides you from CLI install through writing, deploying and calling agent formations

  • 178 ★ on the repository
  • Last commit
Skill

Extended access

Why this level

  • The skill leads the agent to install a server and deploy formations; in the docs the Linux installer runs through `sudo`
  • Formations hold model provider keys: secrets belong in the platform mechanism, not in repository files
All reasons and checks

muxi-ai/muxiAuthor:muxi-ai

Install

In your terminal, with SkillFoxx CLI

npx skillfoxx add skills/muxi

Detects the agents on your machine, shows the access level and pins the version.

Other ways to install

Access and checks

How we review

Extended accessThe tool runs code or makes network calls and only reads external data.

  • The skill leads the agent to install a server and deploy formations; in the docs the Linux installer runs through `sudo`
  • Formations hold model provider keys: secrets belong in the platform mechanism, not in repository files

This is third-party code. Review the repository files before installing.

What it does

MUXI is an open-source server for running AI agents in production: a Go server, a Python runtime, the `muxi` CLI and 12 SDKs. Agents, memory, tools and knowledge are declared in a "formation", a portable set of `.afs` files deployed with `muxi deploy`. The skill sits in the repository's `skills/muxi` directory and gives the agent a reference for the platform. It covers CLI and server installation, configuration, CLI commands, secrets, writing formations in the Agent Formation Schema, deployment, registry operations and both APIs, the Server API and the Formation API. Detailed tables live in `references/`: commands, formation schema, examples, SDKs. The agent loads them on demand. The skill activates on MUXI questions such as "how do I deploy a formation", "where do secrets go", "how do I attach an MCP tool". The minimal flow from the docs is `muxi pull @muxi/hello-muxi`, then `muxi deploy` and `muxi chat hello-muxi`.

Who it is for. Developers deploying their own agents on MUXI who want Claude Code or Cursor to write formations and deploy commands themselves.

Good fit when

  • First formation: the agent installs the CLI, pulls `@muxi/hello-muxi` and checks the answer with `muxi chat`
  • You write `.afs` files from scratch and want the formation schema at hand instead of recalled from memory
  • Secrets and model provider keys must be laid out the way the platform expects
  • A Python or TypeScript app talks to a formation through the Formation API or an SDK
  • MCP tools are attached to a formation and you need a working configuration example

Not a fit when

  • The agent is built on LangChain, CrewAI or another framework: the skill covers MUXI only
  • You need the internals of the server or the runtime: the skill describes running the platform and does not walk through server code
  • A single model call without orchestration or memory is enough: a server is overkill for that

Example request

Deploy the hello-muxi formation locally, add a search tool to it and show how to call it from Python through the SDK.

Limitations

The skill is a reference: it does not install MUXI, and the CLI and server are installed separately with the commands from the docs.

The connect command `claude mcp add-skill skills/muxi` is described in `skills/README.md` and runs from the root of a cloned repository. For Cursor the skill is symlinked into `.cursor/skills/`.

The MUXI server and runtime ship under the Elastic License 2.0: commercial use is free, but selling access to MUXI itself as SaaS is not allowed. The CLI, SDKs and schemas are Apache 2.0.

The performance figures in the skill text come from the author. The repository's last commit is from September 2026.

How to disable. Remove the symlink or the `skills/muxi` entry from the agent settings. Deployed formations stay on the MUXI server and are removed with CLI commands.

FAQ

What is a formation in MUXI?

A formation is a portable set of `.afs` files: agents, knowledge, memory, tools, skills, workflows and policies in one deployable package. The skill teaches the agent to write such files against the Agent Formation Schema and to check them with CLI commands.

How do I install the MUXI skill in Claude Code?

Clone the repository and run `claude mcp add-skill skills/muxi` from its root. Or add a `skills` key with the value `["skills/muxi"]` to `.claude/settings.json`. Both ways are described in `skills/README.md`.

How is MUXI different from LangChain?

LangChain is a library, and deployment is code the developer writes. MUXI is a server: agents are declared in files, and orchestration, memory and launch sit in the infrastructure. A comparison table is in the skill itself.

Which SDKs does MUXI have?

The skill names Go, Python and TypeScript and lists nine more as planned, while the repository text mentions twelve SDKs. The current list is in `references/sdks.md` inside the skill.

What license is MUXI under?

The server and runtime ship under the Elastic License 2.0, the CLI, SDKs and schemas under Apache 2.0. Commercial use is free, but selling access to MUXI itself as SaaS is not allowed.

Discussion

No questions or reviews yet.

Skills

An MCP gateway imports servers and skills once and gives Claude Code, Codex and OpenCode short summaries instead of full schemas.

MCP serverBroad access91

A CLI installer and package manager for coding agent skills: one command installs skills into many runtimes and helps build your own

CLIExtended access1.2K

A CLI to run several coding agents: one config, a session console, local memory and a set of engineering skills

CLIBroad access1.7K

Meta Skill (ms)

Meta Skill

Meta Skill (ms) keeps skills locally in SQLite and Git, searches them with a hybrid index and serves agents over MCP

CLIExtended access213

README in short

The README presents MUXI as an AI application server: formations deploy with three commands and are reachable over REST, SSE, MCP and SDKs. The Agent Formation Standard is published as a separate open specification. Skills live in `skills/` and follow the open Agent Skills specification.

README badge

Original SKILL.md

---
name: muxi
description: >
  Guide users through the MUXI platform -- infrastructure for AI agents.
  Covers installation (CLI and server), server setup and configuration,
  CLI commands and workflows, secrets management, writing formations
  (Agent Formation Schema), deploying formations, registry operations,
  and using both the Server API and Formation API.
  Use when the user asks about MUXI setup, CLI commands, formation authoring,
  secrets, deployment, the registry, server configuration, agents, MCP tools,
  overlord, memory, or any "how do I..." question about MUXI.
license: Apache-2.0
metadata:
  author: muxi-ai
  version: "1.0.0"
---

# MUXI Platform

MUXI (Multiplexed eXtensible Intelligence, pronounced /muk-see/) is open-source production infrastructure purpose-built for AI agents. Not a framework. Not a wrapper. A server.

**Core philosophy:** Agents are native primitives -- declared in portable `.afs` files, orchestrated at the infrastructure layer, scaled like containers. No frameworks to fight. No queues to wrangle. Just infrastructure that understands what agents do.

Think: Flask is a framework. Nginx is infrastructure. MUXI is the Nginx for agents.

| | MUXI | LangChain / CrewAI |
|---|---|---|
| **Type** | Server infrastructure | Python library |
| **Deployment** | `muxi deploy` | Write deployment code |
| **Configuration** | Declarative `.afs` files | Imperative code |
| **Multi-tenancy** | Built-in isolation | Build yourself |
| **Observability** | 356 event types, 10+ export targets | Add external tools |
| **Async/Scheduling** | Native support | Add Celery/etc. |

**Key stats:** <100ms avg response, 88.9% test coverage, 92% semantic cache hit rate, 21 LLM providers / 300+ models via OneLLM.

**Licensing:** Server & Runtime = Elastic License 2.0 (free for commercial use, cannot offer MUXI itself as SaaS). CLI, SDKs, Formations, Schemas = Apache 2.0.

## Architecture

| Component | Analogy | Purpose | Language |
|-----------|---------|---------|----------|
| **Server** | Docker engine | Orchestration, routing, lifecycle | Go |
| **Runtime** | Docker images | Formation execution (SIF containers) | Python |
| **CLI** | CLI | Management and deployment | Go |
| **Registry** | Docker Hub | Distribution and discovery | PHP |
| **SDKs** | Client libs | Go, Python, TypeScript (+ 9 planned) | Various |

```
Your Application

Report a problem