Nexus MCP

v0.11.0Coding Agentsstable

MCP server that lets AI models invoke CLI agents (Gemini, Codex, Claude, OpenCode) as tools — with parallel execution, retries, and structured output parsing.

aiapilarge-language-modelsllmopsllms
Share:
428
Stars
0
Downloads
0
Weekly
0/5

What is Nexus MCP?

Nexus MCP is a Model Context Protocol (MCP) server that allows AI assistants like Claude, Cursor, and VS Code to mcp server that lets ai models invoke cli agents (gemini, codex, claude, opencode) as tools — with parallel execution, retries, and structured output parsing.

MCP server that lets AI models invoke CLI agents (Gemini, Codex, Claude, OpenCode) as tools — with parallel execution, retries, and structured output parsing.

This server falls under the Coding Agents category on MCPgee, the world's largest MCP server directory with 33,000+ servers.

Features

  • MCP server that lets AI models invoke CLI agents (Gemini, Co

Use Cases

CLI agent invocation
Parallel execution and retries
Structured output parsing
j7an

Maintainer

LicenseMIT License
Languagerust
Versionv0.11.0
UpdatedMay 19, 2026
Statushealthy
Maintenanceactive

Works with

ClaudeOpenAIwindowsmacoslinux

Installation

NPM

npx -y @createveai/nexus-mcp-server

PIP

pip install nexus-mcp

Manual Installation

npx -y @createveai/nexus-mcp-server

Configuration

Configuration Details

Config File

claude_desktop_config.json

Performance

Response Metrics

Response Time< 200ms
ThroughputMedium

Resource Usage

Memory UsageLow
CPU UsageLow

How to Set Up and Use Nexus MCP

Nexus MCP is a Python-based MCP server that lets AI models invoke CLI coding agents — currently Claude Code, Codex (OpenAI), and OpenCode — as first-class MCP tools, with support for parallel batch execution, automatic retries with exponential backoff, configurable output limits, and structured output parsing. It exposes two core tools (prompt and batch_prompt), ten discoverable MCP prompt templates covering code review, debugging, refactoring, test generation, and model comparison, and read-only resource endpoints for runner availability and configuration. Platform engineers and teams use it to build agent-of-agents workflows where one AI model orchestrates multiple specialist CLI agents in parallel.

Prerequisites

  • Python 3.13+ and the UV package manager installed
  • At least one supported CLI agent installed: Claude Code ('claude' on PATH), Codex ('codex' on PATH), or OpenCode ('opencode' on PATH)
  • API credentials configured for the CLI agents you want to use (e.g., ANTHROPIC_API_KEY for Claude Code, OPENAI_API_KEY for Codex)
  • An MCP-compatible client such as Claude Desktop or Cursor
1

Install Nexus MCP

Install via pip or use uvx for a zero-install run. The recommended approach is uvx which handles isolation automatically.

# Recommended (no install needed):
uvx nexus-mcp

# Or install permanently:
pip install nexus-mcp
2

Install required CLI agents

Install the CLI agents you want Nexus to invoke. Claude Code and Codex are the most common choices.

# Claude Code:
npm install -g @anthropic-ai/claude-code

# Codex:
npm install -g @openai/codex
3

Add Nexus MCP to your client config

Register the Nexus MCP server with Claude Desktop or another MCP client. Use the npx variant from the npm package or the uvx variant from PyPI.

{
  "mcpServers": {
    "nexus": {
      "command": "npx",
      "args": ["-y", "@createveai/nexus-mcp-server"]
    }
  }
}
4

Configure optional environment variables

Set Nexus environment variables to control timeouts, output limits, and retry behaviour. Per-agent overrides use the NEXUS_{AGENT}_{KEY} pattern.

export NEXUS_TIMEOUT_SECONDS=600
export NEXUS_RETRY_MAX_ATTEMPTS=3
export NEXUS_EXECUTION_MODE=default
# Per-agent example:
export NEXUS_CLAUDE_MAX_RETRIES=5
5

Verify runner detection

Ask your MCP client to read the nexus://runners resource to confirm which CLI agents were detected at startup.

Nexus MCP Examples

Client configuration

Claude Desktop config using the npm package. The PyPI package can also be used with 'uvx nexus-mcp' as the command.

{
  "mcpServers": {
    "nexus": {
      "command": "npx",
      "args": ["-y", "@createveai/nexus-mcp-server"]
    }
  }
}

Prompts to try

Example prompts for using the Nexus MCP tools from within an MCP client.

- "Use the batch_prompt tool to ask both Claude Code and Codex to review src/auth.py for security issues in parallel."
- "Run the code_review prompt template on src/api/routes.py with focus on error handling."
- "Use the compare_models prompt to ask Claude Code and Codex the same refactoring question and show me both answers."
- "Debug the error 'AttributeError: NoneType has no attribute split' in src/parser.py using the debug prompt template."
- "Generate unit tests for src/utils.py using pytest with 90% coverage goal."

Troubleshooting Nexus MCP

No runners are detected at startup

Nexus detects CLI agents by running them on PATH during startup (with a 30-second timeout). Ensure 'claude', 'codex', or 'opencode' are installed globally and accessible on the PATH that the MCP server process inherits. Check with 'which claude' or 'which codex' in the same shell.

batch_prompt tasks time out or return partial output

Increase NEXUS_TIMEOUT_SECONDS and NEXUS_TOOL_TIMEOUT_SECONDS. For very large outputs, increase NEXUS_OUTPUT_LIMIT_BYTES — output beyond the limit spills to a temp file referenced in the response.

Retries are not triggering on transient failures

Set NEXUS_RETRY_MAX_ATTEMPTS to a value greater than 1 (default is 3). Check NEXUS_RETRY_BASE_DELAY and NEXUS_RETRY_MAX_DELAY to tune the exponential backoff window.

Frequently Asked Questions about Nexus MCP

What is Nexus MCP?

Nexus MCP is a Model Context Protocol (MCP) server that mcp server that lets ai models invoke cli agents (gemini, codex, claude, opencode) as tools — with parallel execution, retries, and structured output parsing. It connects AI assistants to external tools and data sources through a standardized interface.

How do I install Nexus MCP?

Install via npm with the command: npx -y @createveai/nexus-mcp-server. Then add the server configuration to your AI client's JSON config file (e.g., claude_desktop_config.json or .cursor/mcp.json).

Which AI clients work with Nexus MCP?

Nexus MCP works with all major MCP-compatible AI clients including Claude Desktop, Claude Code, Cursor, VS Code (GitHub Copilot), Windsurf, and Cline.

Is Nexus MCP free to use?

Yes, Nexus MCP is open source and available under the MIT License license. You can use it freely in both personal and commercial projects.

Browse More Coding Agents MCP Servers

Explore all coding agents servers available in the MCPgee directory. Each server includes setup guides for Claude, Cursor, and VS Code.

Quick Config Preview

{ "mcpServers": { "nexus": { "command": "npx", "args": ["-y", "@createveai/nexus-mcp-server"] } } }

Add this to your claude_desktop_config.json or .cursor/mcp.json

Read the full setup guide →

Ready to use Nexus MCP?

Browse our complete directory of 33,000+ MCP servers, read setup guides for your editor, and start building with the Model Context Protocol.

33,000+ ServersFree & Open SourceStep-by-Step Guides