The coordination layer for autonomous agents

Build agents that
move together.

Mushi is the multi-chain agentic coordination layer for building, coordinating, and running autonomous AI agents across Solana, Robinhood Chain, Base, and BSC.

Launching on PangeaSolana·Robinhood Chain·Base·BSC
01 — The framework

Everything you need to
ship autonomous agents.

One open, server-agnostic stack for agents that think, coordinate, and execute natively across multiple chains.

01

Testing Framework

Playwright E2E tests and live simulations across real multi-chain agent runtimes.

Test coordination paths, retry behavior, and cross-chain execution in deterministic environments before shipping.
02

Mushi Inspector

Inspect autonomous agents in replicated blockchain environments before shipping.

Trace decisions, inspect shared context, and replay every action across an isolated, observable chain environment.
03

Multi-Chain Dev Server

Hot reloading agent server with deterministic chain data and local forks.

Develop against Solana, Robinhood Chain, Base, and BSC from one local runtime with fast feedback loops.
04

Agent UI Components

Pre-built components for agent dashboards, tasks, actions, and onchain state.

Compose premium interfaces for agent status, approvals, execution history, and shared onchain state.
05

Agent APIs

20+ typed React hooks for coordination, state, wallets, and cross-chain execution.

Use typed primitives for orchestration, identity, wallets, messaging, and reliable multi-chain actions.
06

CLI Tools

Scaffold, build, test, simulate, and ship agents from one focused command line.

Move from a blank project to a production-ready agent with one consistent developer workflow.
02 — Convention over configuration

From idea to
coordination.

Agents should spend their time solving problems, not wiring infrastructure. Mushi gives every project a clear path from local simulation to multi-chain execution.

import { createAgent } from 'mushi'

export default createAgent({
  name: 'treasury-agent',
  chains: ['solana', 'base', 'bsc'],
  tools: ['rebalance', 'route'],
  coordinate: true,
})
npx mushi dev ● running locally
03 — Built for what's next

Agents are better
when they coordinate.

Single-chain agents are useful. A network of agents that can reason across ecosystems, share context, and act together is a new primitive.

01

One codebase.
Every chain.

Build once and deploy across Solana, Robinhood Chain, Base, BSC, and whatever comes next.

02

Coordination
by default.

Give agents shared state, clear roles, and the tools to make decisions together.

03

Simulate before
you ship.

Test every path in deterministic environments before your agents touch real value.

04 — Start here

Popular docs

05 — Questions, answered

Frequently
asked.

What is Mushi?

Mushi is the Multi-Chain Agentic Coordination Layer for building, coordinating, and running autonomous AI agents across multiple blockchains.

Which chains does Mushi support?

Mushi is designed for Solana, Robinhood Chain, Base, and BSC from day one.

What is the $MUSHI token?

$MUSHI launches on Pangea.trade with a 2% trading fee and is paired across all four chains from day one.

Is Mushi open source?

Mushi is built to be composable, observable, and accessible to the teams building the next generation of onchain software.

How do I start building?

Mushi is built to be composable, observable, and accessible to the teams building the next generation of onchain software.