ZCode Review 2026: Free AI Coding Agent With Goal Mode (vs Cursor)

I wanted a free desktop coding agent that could chew through long tasks without me babysitting it. That is basically ZCode: download it, hook up a model, and let Goal Mode keep going until the job is done. Bot Channels put the same session on your phone. Multi-model is supported if you do not want to live on GLM alone.

ZCode is the official desktop app from Z.AI, the folks behind GLM. File manager, terminal, Git, browser preview, and agent chat sit in one window. Open a project, tell it what you need, and it reads files, writes code, runs commands, and commits.

I have been on it since launch, and the pace has not slowed: the app is now at v3.15.3, Linux is fully supported, and the whole harness was open-sourced at github.com/zai-org/ZCode. Cursor and Windsurf still feel like editors with chat stapled on. ZCode is the opposite: the agent is the product, the UI just follows. Z.AI calls it an “Agentic Development Environment,” which is marketing language, but it matches how the app actually behaves.

What ZCode does

You interact through natural language, and the agent handles file operations, terminal commands, and browser previews. Here’s the workspace you see when you open it:

Installation

ZCode is available as a desktop app for macOS (Apple Silicon and Intel), Windows (x64 and ARM64), and Linux.

macOS

1. Download the DMG from zcode.z.ai
2. Open the DMG file
3. Drag ZCode.app to your Applications folder
4. Launch from Launchpad

Windows

1. Download the installer from zcode.z.ai
2. Double-click the installer
3. Follow the setup wizard

Linux

Linux is fully supported now — pick between AppImage, DEB, and RPM builds for x64 and ARM64 on the download page. Closing the window keeps the app in the system tray, which is what lets scheduled and idle-time tasks keep running in the background.

Download ZCode

Configuring AI providers

ZCode supports multiple providers. You’re not locked into a single service.

Supported providers

Provider	Models Available	How to Connect
Z.AI / BigModel	GLM-5.3, GLM-5.3-Flash, GLM-5.2	Built-in, sign in with your Z.AI or Zhipu account
OpenCode Go	Go’s curated model roster	Provider template added in v3.12.3, paste your Go key
Anthropic	Claude Sonnet, Opus, etc.	Enter Anthropic API key
OpenRouter	100+ models	Set base URL to https://openrouter.ai/api, add API key
Moonshot	Kimi K3, Kimi K2.x	Base URL: https://api.moonshot.cn/anthropic
DeepSeek	DeepSeek Chat	Base URL: https://api.deepseek.com/anthropic
Custom	Any OpenAI/Anthropic-compatible API	Add provider name, base URL, and API key

ZCode Agent

The agent is the core of the experience. ZCode Agent is built in-house and deeply adapted for GLM-5.3. It handles complex project understanding, long-task planning, and continuous code changes across multiple files.

Adding context

Click the + button at the bottom-left of the chat input to add context:

• Attachments: Upload screenshots, documents, and requirement material
• @ mentions: Reference files in the workspace so the agent can locate relevant code
• # conversations: Link a past conversation to bring its context into the current task
• / commands: Call saved prompts to reuse a fixed workflow
• $$ skills: Invoke reusable skill playbooks

Project instructions

ZCode reads an AGENTS.md file from your project root and loads it into every conversation. Put your project conventions, architecture notes, and coding standards here.

Two locations are supported:

Source	Path	Scope
User global	~/.zcode/AGENTS.md	All projects
Workspace	AGENTS.md in project root	Current project

Execution modes

Five modes control how much freedom the agent gets. Press Shift + Tab while the chat input is focused to cycle through them.

Mode	What Happens	Good For
Default Mode	Standard task strategy with balanced confirmations	Everyday development, routine edits
Confirm Before Changes	Asks before every file edit or command	Critical code, production configs
Auto Edit	Edits files automatically, commands still need confirmation	Routine iteration with fewer interruptions
Plan Mode	Creates a plan first, then implements after approval	Complex multi-step tasks
Full Access	Minimal interruptions, agent proceeds continuously	Low-risk tasks where you want speed

Goal Mode

This is one of the newer features and I’ve found it genuinely useful for long tasks. Type /goal <objective> and the agent keeps iterating toward the goal. After each round, it verifies whether the goal is met. If not, it continues automatically — no need to type “continue” every few minutes.

Good use cases for Goal Mode:

• “Refactor the whole module and keep the tests passing”
• “Fix all TypeScript compile errors”
• “Get this page’s Lighthouse performance score above 90”

The summary panel in the top-right shows goal status, elapsed time, total tokens, and iteration count. You can pause, replace, or clear the goal at any time.

For long goal-driven tasks, pair Goal Mode with Full Access or Auto Edit to minimize interruptions and keep iterations flowing.

Bot Channel

Bot Channel connects WeChat and Feishu to your ZCode workspace. After setup, you can message the bot from your phone’s chat app to check task progress, send instructions, and keep the agent moving.

Setting up Feishu:

1. Open Bots settings in ZCode’s left sidebar
2. Click Create Bot and select Feishu
3. Scan the QR code — ZCode creates the Feishu app automatically
4. Send /bind pairing-code in the Feishu conversation
5. The bot is ready to use

WeChat works similarly — scan to sign in and bind automatically. Since v3.15 the bot can also collaborate with you inside Feishu group chats and standalone topics. DingTalk, Discord, and WeCom support are still on the roadmap.

Remote Control

Remote Control lets your phone connect to the current desktop workspace. Scan a QR code, and your phone shows a live view of the session. You can read agent responses, send new instructions, and confirm or stop long-running tasks.

The desktop stays as the runtime — the phone only handles display and text input. One phone connection per session. For longer-lived access from chat apps, use Bot Channel instead.

Skills system

Skills are Markdown files that teach the agent how to handle specific tasks. They work like reusable playbooks.

Type	Storage	Scope
User Skills	~/.claude/skills/	All projects
Plugin Skills	Installed via plugins	Auto-available
Project Skills	.claude/skills/	Current project only

MCP services

ZCode uses the Model Context Protocol to extend what agents can do. Three built-in MCP services cover common needs:

Plugin system

Plugins add new commands, agents, MCP connections, language servers, and more. ZCode ships with access to the Claude Code plugin marketplace, so you can install community plugins you’re already familiar with.

Plugin Type	What It Adds
Agent	Specialized agents with domain workflows
Command	Custom / slash commands
MCP	External tool and data source connections
LSP	Language servers for code navigation and type checking
Skill	Behavior guidelines for the agent
Hook	Auto-execute actions on specific events

Version control

ZCode handles versioning at two levels.

Conversation-level rollback: Every message creates a checkpoint. Review all file changes after any interaction in a multi-file diff view, undo only the last interaction’s changes, or jump back to the state after any specific message.

Built-in Git panel: View modified files with status markers, write commit messages and commit with one click, switch branches, and browse commit history.

ZCode vs other AI coding tools

Feature	ZCode	Cursor	Claude Code CLI	Windsurf
Type	Standalone ADE	IDE (VS Code fork)	Terminal CLI	IDE (VS Code fork)
Open source	Yes (zai-org/ZCode)	No	Yes	No
Official Z.AI tool	Yes	No	No	No
Multi-Model	GLM, Claude, GPT, Kimi, DeepSeek, OpenCode Go, custom	Multiple providers	Claude only	Multiple providers
Goal Mode	Yes, auto-iterate until verified	No	No	No
Workflows	Multi-sub-agent orchestration via /workflow	No	Subagents	No
Bot Channel	WeChat, Feishu (incl. group chats)	No	No	No
Skills System	Yes, reusable Markdown playbooks	No	No	No
Plugin Marketplace	Agents, commands, MCP, LSP, UI plugins	Extensions	No	Extensions
Chat Versioning	Conversation checkpoints + rollback	No	No	No
Remote Dev (Mobile)	QR code + Bot Channel	No	No	No
AGENTS.md	Auto-loaded per project	.cursorrules	CLAUDE.md	.windsurfrules

Getting started in 5 minutes

Step 1: Download and install ZCode from zcode.z.ai

Step 2: Open the app and click “Connect” to set up your AI provider. If you have a GLM Coding Plan, bind it for the best experience with GLM-5.3 and the 1.5x quota benefit.

Step 3: Open a project folder in ZCode.

Step 4: Type /init or create an AGENTS.md file in your project root with your conventions.

Step 5: Start chatting. Ask the agent to explain your codebase, fix a bug, or build a new feature.

Get Started with ZCode

GLM Coding Plan pricing

Z.AI’s GLM Coding Plan was relaunched in 2026 with a credit system: Lite, Pro, and Max tiers (Pro gives 6x the Lite allowance, Max 14x), starting at $18/month. Every plan has a 5-hour credit pool that refills on a rolling basis plus a weekly cap that resets every 7 days. Off-peak usage charges half the standard credit rate, so timing your sessions stretches the plan a lot.

Get GLM Coding Plan (10% OFF)

Pros and cons

What I like:

• GLM-5.3 integration is deep — the agent, model, and execution workflow are tuned together
• Goal Mode saves me from typing “continue” every 5 minutes on long tasks
• Workflows orchestrate multiple sub-agents with per-agent models — the biggest addition since Goal Mode
• Bot Channel means I can check on tasks from WeChat without opening my laptop
• The 1.5x quota benefit with GLM Coding Plans makes it cheaper to run than alternatives
• Full Linux support (AppImage, DEB, RPM) — wasn’t there at launch
• The whole harness is open source now

What I don’t:

• The plugin ecosystem is still smaller than Cursor’s extension marketplace, though UI plugins (MCP Apps) arrived in v3.15
• Some MCP services need Zhipu/BigModel API keys specifically
• The learning curve is real — agents, skills, plugins, MCP, Goal Mode, workflows, execution modes. There’s a lot to figure out.
• Bot Channel only supports WeChat and Feishu right now (Discord and Slack are still promised)

Frequently asked questions

Is ZCode free?

ZCode itself is free and open source. You pay for the AI models you connect — either through Z.AI’s GLM Coding Plans, OpenRouter credits, Anthropic API usage, or whichever provider you choose. New users get a 5-day free trial with 8M tokens/day.

Do I need Claude Code to use ZCode?

No. ZCode supports multiple AI providers. You can use GLM-5.3, models through OpenRouter or OpenCode Go, the Anthropic API directly, Moonshot (Kimi), DeepSeek, or any service that speaks the OpenAI or Anthropic protocol.

What is the GLM Coding Plan quota benefit?

GLM Coding Plan subscribers using ZCode get roughly 1.5x the effective usage compared to other supported tools. On the relaunched plans, usage is metered in credits with a rolling 5-hour pool and a weekly cap; off-peak hours charge 0.5x the standard credit rate. The same plan simply lasts longer in ZCode.

Does ZCode work on Linux?

Yes, Linux is fully supported. AppImage, DEB, and RPM builds are available for x64 and ARM64 on the download page, and closing the window keeps the app in the system tray so background tasks keep running.

Can I use ZCode with local models?

Yes. ZCode supports any OpenAI or Anthropic-compatible API endpoint. Point it at a local inference server (like Ollama or vLLM) and it’ll work with whatever model you’re running.

How does ZCode compare to Cursor?

Cursor is a VS Code fork with AI editing features. ZCode is a standalone environment built around agent interactions — and it’s open source. ZCode has Goal Mode, Bot Channel, multi-agent workflows, specialized agents, a skills system, and conversation-level version rollback. Cursor has a more familiar IDE experience and a larger community.

Wrapping up

ZCode has come a long way since its initial launch. The last few months alone added dynamic workflows, Office and Coding modes, Autonomous Mode, UI plugins, per-workspace plugins, idle-time tasks, an OpenCode Go provider template, and full Linux support — plus the whole harness went open source under zai-org/ZCode. As Z.AI’s official coding environment for GLM-5.3, it gets the deepest integration with their models and the quota benefits that come with it.

The Goal Mode feature alone has changed how I handle long tasks — set it and walk away. Come back to a completed result. The Bot Channel is useful for checking on progress from your phone without needing to be at your desk.

If you’re already using GLM models or want to try them, ZCode is the way to go. The 5-day free trial and 1.5x quota benefit make it worth testing.

Try ZCode Free