YunmengZe Agent CLI
A terminal coding agent in Go that edits files and runs commands in a UI or scriptable CLI and works with any model.
- 228 ★ on the repository
- Last commit
- Apache-2.0
Broad access
Why this level
- In agent mode tools perform file edits and shell commands with your account permissions without confirmations.
- The install scripts run through curl or PowerShell and download a release.
Install
Manual install
brew install --cask yyZe0122/tap/ymzFor Windows `scoop install ymz/ymz`, for Linux and macOS the `install-user.sh` script from the README. Then `ymz version`.
Access and checks
How we reviewBroad accessThe tool can change data, run commands or handle secrets.
- In agent mode tools perform file edits and shell commands with your account permissions without confirmations.
- The install scripts run through curl or PowerShell and download a release.
This is third-party code. Review the repository files before installing.
What it does
`ymz` is a single executable for Windows, Linux and macOS with no separate service. In the terminal you can switch models, resume conversations, steer the active turn and answer agent questions. `plan` mode only inspects the project and discusses: the tool registry blocks file changes, shell, memory and MCP. `agent` mode (the default) executes tools with your OS account access, including file changes and shell commands, without approval prompts. OpenAI-compatible services (Chat Completions, Responses), Anthropic Messages and Gemini are supported, including compatible local endpoints. Sessions, context summaries and memory live in a local SQLite database. File skills and MCP servers load optionally, up to two independent child tasks run, and web search and image analysis work. Each capability can be disabled in configuration. Start with `ymz`, `ymz --mode plan`, `ymz --continue`, and in scripts `ymz run --json "Summarize the changes in this workspace"`.
Who it is for. Developers who need a terminal coding agent not tied to one model provider and callable from scripts.
Good fit when
- The agent has to work with OpenAI, Anthropic or a local model, switchable mid-session.
- The structure of an unfamiliar project must be studied in `plan` mode without changing anything.
- A summary of changes is needed from a script in JSON format.
- Skills and MCP servers are needed in a simple single-file terminal agent.
Not a fit when
- You need confirmation before every edit: in `agent` mode none is requested.
- The agent must run in a strictly confined environment: disabling web tools does not restrict networking by the shell.
- You need rich IDE integration: the project is built for the terminal.
Example request
ymz run --mode plan "Explain how this project is organized" and show the key modules.Limitations
In `agent` mode tools work with your account permissions without confirmation prompts, so for unfamiliar projects start with `plan`. Disabling web tools does not restrict networking for shell commands, MCP servers and image URLs. Model calls and enabled web tools go to the services you configured. The install scripts `install-user.sh` and `install.ps1` download the release and verify SHA-256. The license is Apache-2.0, the last commit dates from October 2026.
How to disable. Remove the `ymz` executable the way you installed it (Scoop, Homebrew or the installer) and the `YMZ_HOME` configuration directory with the `core.db` database.
FAQ
How do I install ymz?
On macOS `brew install --cask yyZe0122/tap/ymz`, on Windows through Scoop (`scoop install ymz/ymz`), on Linux and macOS with the `install-user.sh` script. After installing reopen the terminal and run `ymz version`.
How does plan mode differ from agent?
In `plan` the agent inspects and discusses while file edits, shell, memory and MCP are blocked. In `agent` tools execute immediately.
Which models can be connected?
OpenAI-compatible Chat Completions and Responses, Anthropic Messages, Gemini and compatible local services. `ymz models` shows the list.
Where are sessions stored?
In a local SQLite database `core.db` in the configuration directory, which `YMZ_HOME` changes.
How do I check the configuration?
With `ymz config validate`, which does not contact model services.
Discussion
No questions or reviews yet.
Related
CLI toolsA terminal coding agent in Go: GLM, DeepSeek and local Ollama models, 59 built-in tools and MCP
A Python terminal coding agent in the manner of Claude Code: reads the repository, edits files, runs commands, with memory, sub-agents and MCP
A Rust coding agent with a native GUI, a terminal mode, an ACP mode for Zed and an MCP server
A coding agent with a Python core and three front-ends: a script CLI, a terminal TUI and an Electron app, for OpenAI-style APIs
Work modes for AI coding agents: a light solo mode for daily tasks and /man with a plan, checks and review for big changes
Kode CLI is a terminal coding agent that reads AGENTS.md and CLAUDE.md, connects MCP servers and works with many models.
Comparison with similar tools
Cards that share the main search query, and how they differ.
| Tool | Official | Repository stars | Main difference |
|---|---|---|---|
| YunmengZe Agent (this card) | no | 228 | |
| Crush | no | 29,067 | Crush is also a terminal agent, while ymz separates plan and agent modes and keeps memory and sessions in one SQLite database. |
| OpenCode | no | 226,389 | OpenCode is a widely used terminal agent, while ymz ships as one executable with local preference memory. |
README in short
An open terminal coding agent in Go with plan and agent modes, support for skills, MCP, child tasks and several model providers.