# flow-nexus [Health: Active]

**Category:** 🤖 Coding Agents  
**Repository:** https://github.com/ruvnet/flow-nexus  
**GitHub Stars:** 106  
**npm Downloads (last month):** 15932  
**Views:** 1  
**Installs:** 0  
**Upvotes:** 0  
**Directory Page:** https://allmcps.com/mcp/flow-nexus

## Description
Cloud-powered AI platform with multi-agent swarms, sandboxes, and workflow automation

## Claude Desktop Quick Installation
Install path detected from listing signals. Uses `npx` (confidence: high):

```json
"mcpServers": {
  "flow-nexus": {
    "command": "npx",
    "args": ["-y","flow-nexus"]
  }
}
```

## Documentation

## What flow-nexus MCP server does

The flow-nexus MCP server exposes Flow Nexus capabilities through the Model Context Protocol. The platform combines cloud-hosted agent swarms, isolated sandboxes, workflow automation, distributed neural processing, monitoring, and a credit-based usage model. It is intended for MCP-enabled development clients that need to provision or coordinate cloud resources from an agent workflow.

Documented examples include registering and logging into a Flow Nexus account, initializing a swarm with a selected topology and agent limit, and creating a sandbox from a template.

## How it works

The package is distributed through npm and can be started with `npx flow-nexus`. Its MCP entry point is launched with:

```bash
npx flow-nexus@latest mcp start
```

Once connected to an MCP client, agents can call Flow Nexus tools. The documented deployment sequence is to authenticate first, then initialize a cloud swarm and create a sandbox. A swarm can be configured with a topology such as `mesh` and a maximum number of agents. A sandbox can be created from a runtime template, with `node` shown as an example.

The platform describes three connected areas: autonomous agents, agentic sandboxes, and neural processing. It also lists workflow automation with event-driven pipelines and verification. Credit usage is part of the platform model, and the README advertises 256 rUv credits for new users; availability and terms should be confirmed with the service.

## Setup and configuration

Install or run the package through npm rather than a local build:

```bash
npx flow-nexus
```

For Claude Desktop, the README provides this MCP registration command:

```bash
claude mcp add flow-nexus npx flow-nexus@latest mcp start
```

Account registration can be performed from the command line:

```bash
npx flow-nexus@latest auth register -e your@email.com -p secure-password
```

The same operations are shown as MCP calls:

```text
mcp__flow-nexus__user_register({ email: "your@email.com", password: "secure" })
mcp__flow-nexus__user_login({ email: "your@email.com", password: "secure" })
```

It does identify `flow-nexus.ruv.io` as the platform website.

## Tools and capabilities

The flow-nexus MCP server is described as providing more than 100 MCP tools. Supported capability areas named in the material include:

- Registering and authenticating users
- Initializing multi-agent swarms
- Creating isolated cloud sandboxes
- Running Node.js, Python, React, Next.js, and custom-container workloads, as listed by the platform
- Building event-driven workflows with message queues and parallel tasks
- Distributed neural-network processing and WASM acceleration
- Monitoring streams, performance metrics, and health dashboards
- Publishing and deploying community templates
- Participating in coding challenges, tournaments, and leaderboards

The README specifically names `user_register`, `user_login`, `swarm_init`, and `sandbox_create` in its examples.

## Limitations and notes

The documented workflow requires an account before deploying a swarm or sandbox.

The README references Claude Desktop, Claude Code, OpenAI Codex, Cursor, GitHub Copilot, and other MCP-enabled tools. Only Claude Desktop has a specific registration command in the provided setup instructions.

_Full upstream README: https://allmcps.com/mcp/flow-nexus/readme_