Operate Monad

Start with getting started（简体中文：开始使用 Monad）. These guides treat Monad as a daemon-first agent team runtime. Monad Mesh owns the team; agent runtimes — first-party or external — do the work under that authority. Every guide shows the monad command first, because the CLI is the complete, scriptable surface. Where the browser does the same job, the guide names the Studio screen next to the command; the handful of things only the browser can do are listed in the web UI.

Build the mesh

Monad Mesh is the agent team runtime. Read this first: it decides who is on the team and how their work is bound, observed, and approved.

Doc	Covers
from-one-agent-to-team.md	Growing from one supervised agent to a durable multi-agent project
choosing-runtime.md	Which agent runtime should execute a given piece of work
mesh-agents.md	Running provider-native agent runtimes as team members and observing them
native-cli-approvals.md	The per-agent autopilot switch for native CLI agents

Run the first-party agent

Monad Agent Runtime is Monad’s own agent runtime and the default mesh member. These guides configure what it can call and know.

Doc	Covers
model-providers.md	Connect the models it calls
skills.md	`SKILL.md` skills: using, writing, gating, forked execution, management
mcp.md	MCP servers: stdio and HTTP, secrets, OAuth, trust controls
computer-use.md	Computer use and browser use via off-the-shelf MCP servers

Operate the daemon

Doc	Covers
installation.md	System requirements, release installer, manual installation, upgrades, and removal
sessions.md	Sessions and agents: create, branch, restore, watch across clients
cli.md	Every command, global flags, aliases, exit codes, scripting patterns
web.md	The browser client: routes, Studio map, and what only it can do
tui.md	Starting the terminal client and the keys it needs
channels.md	IM channels: connections, group rules, and in-chat commands
api.md	The daemon API: transports, auth, realtime streams, error envelope, OpenAPI

Tune trust

Doc	Covers
sandbox-backends.md	Choosing and configuring sandbox backends
agent-runtime-credentials.md	Write-only credentials for generated code and shell processes
privacy.md	What is stored locally, and what leaves the machine

Maintain

Doc	Covers
troubleshooting.md	Symptom → cause → command, across every subsystem
releases.md	Channels, upgrading, rollback, versioning and compatibility

Read Monad’s architecture to understand why the runtime behaves this way.

Get Started

Monad Mesh: The Agent Team Runtime

Monad Agent Runtime: The First-Party Agent

Operate the Daemon

Trust and Policy

Reference

Operate Monad

Build the mesh

Run the first-party agent

Operate the daemon

Tune trust

Maintain

​Build the mesh

​Run the first-party agent

​Operate the daemon

​Tune trust

​Maintain

Build the mesh

Run the first-party agent

Operate the daemon

Tune trust

Maintain