zodex
infinite token glitch - mcp server for chatgpt to use codex tools on a real machine
Install / Use
claude mcp add amxv -- npx -y github:amxv/zodexIf the server publishes to npm under a different name, use that package instead — check the repo README.
MCP Server
Model Context Protocol server
Quality Score
Category
AI & Machine LearningSupported Platforms
Our assessment of zodex
zodex scores 75/100 on our quality scale, 817th of 947 AI & Machine Learning skills we index.
Its MCP Server is 5.9 KB long, split into 7 sections with 8 code examples: a solid amount of guidance for an agent.
It has 10 GitHub stars, so there is little community track record yet; judge it on its content.
Maintenance, license and trust
- The repository was last updated today, so zodex is actively maintained.
- Our last check on 2026-11-10 found the source still online.
- It is released under the Apache-2.0 license, a permissive license that allows use, modification and commercial use with attribution.
- Its trust signals score 92/100, with 1 caution from licensing, adoption, age or documentation. These come from repository metadata, not a code audit — read the skill file before letting an agent act on it.
Safety scan
No issues foundOur scan of the whole file found no instruction hijacking, hidden characters, credential access, data exfiltration or destructive commands (2 minor notes below). An AI review of the same text found nothing harmful.
- noteInstalls by piping a downloaded script into a shellline 34
curl -fsSL https://zodex.ashray.xyz/install.sh | bash - noteInstalls by piping a downloaded script into a shellline 41
irm https://zodex.ashray.xyz/install.ps1 | iex
AI review by kimi-k2.7-code on 2026-11-10. Automated pattern scan on 2026-11-10. It catches known dangerous patterns, not every risk — read a skill before letting an agent act on it.
zodex compared with similar skills
All 4 of these similar skills score higher than zodex; compare them before choosing.
| Skill | Score | Stars | Updated | Format |
|---|---|---|---|---|
| zodex (this skill)by amxv | 75 | 10 | today | MCP Server |
| claude-memby thedotmack | 100 | 117.6k | today | CLAUDE.md |
| Agent-Reachby Panniantong | 100 | 112.3k | 8d ago | CLAUDE.md |
| Understand-Anythingby Egonex-AI | 100 | 92.8k | 1d ago | CLAUDE.md |
| rufloby ruvnet | 100 | 81.9k | today | CLAUDE.md |
Frequently asked questions
- How do I install zodex?
- Run
claude mcp add amxv -- npx -y github:amxv/zodex. The install tabs above show the steps for each supported agent. - Which AI agents does zodex work with?
- It is written for Claude Code, Claude Desktop and OpenAI Codex, as a MCP Server file. Other agents that read the same format can often use it too.
- Is zodex safe to use?
- Our scan of the whole file found no instruction hijacking, hidden characters, credential access, data exfiltration or destructive commands (2 minor notes below). An AI review of the same text found nothing harmful. It is Apache-2.0-licensed and scores 92/100 on trust signals. Skills are instructions an agent will follow, so read the file before installing it and do not approve commands you do not understand.
- Is zodex still maintained?
- The repository was last updated today, so zodex is actively maintained.
Skill content
View source on GitHubzodex
Give ChatGPT a real terminal on a real machine.
Zodex is a small MCP coding harness that gives ChatGPT the command, stdin, and patch primitives coding agents already know—without wrapping your development environment in a new abstraction.
Run it directly on a trusted Apple Silicon Mac, 64-bit Linux machine, or x86_64 Windows machine with Zodex Local, or use the same core execution tools on a wake-on-demand remote Linux workspace with Zodex Sprite.
<p align="center"> <img src=".github/assets/zodex-local-liveboard.png" alt="ChatGPT coding on a Mac through Zodex Local, with the Zodex Liveboard showing commands and file changes beside it" width="100%"> </p>Zodex Local
Local puts ChatGPT on your trusted Linux, macOS, or Windows host through an OpenAI Secure MCP Tunnel. There is no public inbound port, proxy server, container, or separate remote machine between the model and the development environment you already use.
One Local runtime can serve multiple ChatGPT conversations at once. Zodex gives each conversation its own short Agent identity, keeps durable execution history, and lets you watch every command and file change in real time through the cross-platform browser Liveboard or terminal viewer.
- Your real environment — shell, Git, language toolchains, credentials, and files run as your logged-in user. Linux and macOS use the captured login shell; Windows Local executes commands through Windows PowerShell with the captured user environment.
- Four Local MCP tools — the three shared execution tools (
exec_command,write_stdin,apply_patch) plus optionalset_agent_aliasfor Liveboard naming. - Many Agents, one runtime — multiple ChatGPT conversations can work concurrently across repos and worktrees.
- Live observability — commands, output, diffs, process state, and Agent timelines in the browser Liveboard or terminal TUI.
- Durable history — inspect previous tool activity even after the viewer or Local runtime stops.
- Explicit lifecycle — start access when you want it, optionally give it a TTL, and stop it cleanly.
- Host-native controls — Linux, macOS, and Windows include Liveboard and a terminal viewer. macOS has a native Swift menu bar app; Windows and graphical Linux desktops have a lightweight Rust system tray. The menu/tray starts at login by default while Local remains off, with a separate opt-in Launch Tunnel at Login checkbox and
zodex local autostartterminal commands. Linux stores the tunnel key in a user-only credential file and Windows uses Windows Credential Manager.
Local is intentionally a trusted-host mode, not a sandbox. ChatGPT receives the same filesystem and developer-tool access your logged-in user has, subject to the host operating system's normal permissions.
Quick start
Install the operator CLI on macOS/Linux:
curl -fsSL https://zodex.ashray.xyz/install.sh | bash
export PATH="$HOME/.local/bin:$PATH"
On Windows PowerShell:
irm https://zodex.ashray.xyz/install.ps1 | iex
Set up the Secure MCP Tunnel once:
zodex local setup
Then start Zodex from the repo you want ChatGPT to see first:
cd ~/code/my-project
zodex local start --ttl 4h
Watch the Agents work. Linux, macOS, and Windows open Liveboard by default:
zodex local watch
And turn access off when you're done:
zodex local stop
See the Local quick start for tunnel setup, ChatGPT connection, menu-bar controls, history, configuration, and troubleshooting.
Two execution modes
The core execution interface stays the same. You choose where it runs and which trust model you want. Local also provides agent-managed Liveboard aliases.
| | Local | Sprite | | --- | --- | --- | | Host | Your Apple Silicon Mac, x86_64/aarch64 Linux machine, or x86_64 Windows machine | Wake-on-demand remote Linux Sprite | | Connection | OpenAI Secure MCP Tunnel | Cloudflare Worker → Sprite | | Trust model | Trusted host; your logged-in user | Restricted Agent account + isolated GitHub writer boundary | | Lifecycle | Explicit start/stop, optional TTL | Automatic wake/sleep | | GitHub access | Your existing local credentials | Review-first PRs, temporary push grants, or scoped YOLO | | Observability | Linux/macOS/Windows Liveboard + TUI; durable history + HTTP/SSE API | Sprite service and operator diagnostics |
For an isolated remote workspace instead of trusted-host execution:
zodex sprite setup --help
See the Sprite quick start.
The entire MCP surface
ChatGPT sees three execution tools in both modes:
exec_command run a command in an explicit absolute workdir
write_stdin poll, interact with, or stop a running process
apply_patch edit files with the Codex-style patch format
Zodex Local only also offers set_agent_alias to name the current conversation's Liveboard column, for example Rindle implementation. This optional tool preserves the Agent's four-character identity, doesn't enter the command timeline, and pushes alias updates to every open Liveboard.
That is deliberately the whole abstraction. Zodex handles the runtime, interactive command I/O, process lifecycle, transport, observation, and host-specific security boundaries underneath it.
Documentation
Full documentation lives at zodex.ashray.xyz/docs.
- Local — trusted Linux/macOS/Windows execution, setup, observability, history, and daily use
- Sprite — remote Linux, GitHub permissions, connection, and operations
- Architecture — how the two execution modes fit together
- MCP tools — exact tool contracts and session behavior
License
Related Skills
claude-mem
117.6kPersistent Context Across Sessions for Every Agent – Captures everything your agent does during sessions, compresses it with AI, and injects relevant context back into future sessions. Works with Claude Code, OpenClaw, Codex, Gemini, Hermes, Copilot, OpenCode + More
Agent-Reach
112.3kGive your AI agent eyes to see the entire internet. Read & search Twitter, Reddit, YouTube, GitHub, Bilibili, XiaoHongShu — one CLI, zero API fees.
Understand-Anything
92.8kGraphs that teach > graphs that impress. Turn any code into an interactive knowledge graph you can explore, search, and ask questions about. Works with Claude Code, Codex, Cursor, Copilot, Gemini CLI, and more.
ruflo
81.9k🌊 The original agent harness. Deploy intelligent multi-player swarms, coordinate autonomous workflows, and build conversational AI systems. Features adaptive memory, self-learning intelligence, federation, vector RAG integration, and native Claude Code / Codex / Hermes and many more Integrated
Languages
Trust signals
From repository metadata: license, adoption, age and documentation. Not a code audit — see the Safety scan above for what the skill file itself contains.
