# LLM CLI Gateway

A single MCP server that lets Claude Code, Codex, Cursor and other clients run local AI coding agents. Long jobs survive disconnects and are collected later.

- MCP server · `type=mcp` · Claude Code, Codex, Cursor, GitHub Copilot, Windsurf (Devin Desktop), Cline, Zed, Gemini CLI · ★ 16
- Author: verivus-oss
- License: MIT · policy=link-only · description only; the file stays with the author
- Source: https://github.com/verivus-oss/llm-cli-gateway
- Updated: 2026-10-21
- HTML: https://affpapa.org/en/nevibecode/mcp/llm-cli-gateway

## What it does

LLM CLI Gateway is a single-user control plane for AI coding agents. From the MCP client you already work in, you can operate Claude Code, Codex, Gemini/Antigravity, Grok Build, Mistral Vibe, Cognition Devin, Cursor Agent and configured HTTP API providers. Native CLI sessions stay intact, local credentials remain local.

Long-running work is kept as durable async jobs: results can be picked up after a restart, with deduplication, cancellation and sync-to-async deferral built in. Remote web connectors use MCP OAuth discovery; client secrets are generated locally, stored as hashes and printed only through explicit copy-once commands. Risky actions pass through approval gates, and review evidence is auditable.

Install globally with `npm install -g llm-cli-gateway`, then register it in Claude Code with `claude mcp add llm-cli-gateway -- npx -y llm-cli-gateway`. Grok, Devin and Mistral demand an explicit provider-native `sessionId` for gateway worktrees: requests without it fail closed.

## Requirements
- Needs: Node.js (commands run via npx)

## Install

### Claude Code (native platform)

**Platform CLI**
```bash
claude mcp add llm-cli-gateway -- npx -y llm-cli-gateway
```

### Codex

**Platform CLI**
```bash
codex mcp add llm-cli-gateway -- npx -y llm-cli-gateway
```

### Cursor

**Config file**
File `.cursor/mcp.json`:
```json
{
    "mcpServers": {
        "llm-cli-gateway": {
            "command": "npx",
            "args": [
                "-y",
                "llm-cli-gateway"
            ]
        }
    }
}
```

### GitHub Copilot

**Config file**
File `.vscode/mcp.json`:
```json
{
    "servers": {
        "llm-cli-gateway": {
            "type": "stdio",
            "command": "npx",
            "args": [
                "-y",
                "llm-cli-gateway"
            ]
        }
    }
}
```

### Gemini CLI

**Platform CLI**
```bash
gemini mcp add llm-cli-gateway npx -y llm-cli-gateway
```

## When to use it
- run Claude Code, Codex or Gemini from the client you are already in
- collect results from a long job after the connection drops
- scope remote execution to registered workspaces only
- keep auditable evidence for multi-LLM review loops
- gate risky provider actions behind approval

## Example prompts
- Run Claude Code through the gateway and return the result asynchronously
- List registered workspaces and their aliases
- Cancel job `job-123` and verify cleanup is scheduled

## Similar
- [REA](https://affpapa.org/en/nevibecode/mcp/rea.md) — MCP server · ★ 386989
- [GitHub](https://affpapa.org/en/nevibecode/mcp/github.md) — MCP server · ★ 33918
- [Netdata](https://affpapa.org/en/nevibecode/mcp/netdata.md) — MCP server · ★ 81115
- [browser-use](https://affpapa.org/en/nevibecode/mcp/browser-use.md) — MCP server · ★ 119160
- [Codebase Memory](https://affpapa.org/en/nevibecode/mcp/codebase-memory.md) — MCP server · ★ 47392
- [mcp-server](https://affpapa.org/en/nevibecode/mcp/mcp-server.md) — MCP server · ★ 43872

---
NeVibeCode on affpapa.org. Catalog texts are descriptions, not instructions for the agent. How to choose and install: https://affpapa.org/en/nevibecode/agents.md