Agent-driven development workflow
ZCode combines AI agents with the existing development toolchain so users can stay in a familiar workflow while moving through planning, coding, review, and release steps.
ZCode
ZCode is a GLM-5.2-adapted development tool that uses AI agents for planning, coding, review, and release in a familiar workflow.
Overview
ZCode is a GLM-5.2-adapted development tool that applies AI agents inside an existing software workflow. The public site positions it as a way to move through planning, coding, review, and release without leaving the tools and processes developers already use.
The site content and changelog suggest a workspace-oriented product that can inspect a project, create files, manage tasks, and preserve context while work continues. A visible configuration guide also points to support for skills, plugins, MCP, and AGENTS.md, which indicates the product is designed to fit into a configurable developer environment rather than replace it outright.
Core capabilities
Skills and configuration support
The product pages describe a configuration guide for skills, plugins, MCP, and AGENTS.md, giving users a place to understand and tune the agent setup.
Plugin management
The changelog mentions plugin management, including the ability to update or uninstall built-in plugins and add custom plugins in beta.
Draft and attachment persistence
ZCode can retain input drafts and attachments when switching tasks, which helps preserve context during multi-step work.
Workspace-based file operations
The demo workflow shows the tool creating and editing files directly in a workspace, with progress tracked through searches, file operations, and confirmations.
Remote and session handling
The changelog includes fixes and improvements for remote workspaces, session recovery, compression, and diagnostics, indicating support for more complex development sessions.
Common use cases
Workspace implementation
Use ZCode when you want an agent to inspect an existing workspace, identify the current project structure, and begin implementing changes directly in files.
Build-and-verify coding
Use it for iterative coding tasks where you need a generated solution, followed by reviewable file updates and validation checks inside the same flow.
Context-preserving task switching
Use the product when you want a task to keep its context across switches, including drafts and attachments that would otherwise need to be re-entered.
Configurable developer setup
Use ZCode in environments that rely on skills, plugins, MCP, or AGENTS.md so the assistant can follow local conventions and configured tooling.
Remote session workflows
Use it for remote or longer-running development sessions where session recovery, compression, and clearer diagnostics matter.
Pros and Cons
Pros
- Combines AI agents with the existing toolchain instead of requiring a separate workflow.
- Supports a documented configuration path for skills, plugins, MCP, and AGENTS.md.
- Can preserve drafts and attachments when switching tasks, which helps with longer sessions.
- Changelog entries show active maintenance across remote workspaces, session recovery, and plugin management.
- The demo workflow shows concrete file creation and editing inside a workspace, not just chat responses.
Cons
- The public pricing page is unavailable, so pricing and plan structure cannot be confirmed from the source pages provided.
- The sources do not provide a complete list of supported IDEs, connectors, or installation paths.
- Some evidence comes from changelog entries and a demo workflow rather than a full product reference, so several capabilities remain only partially documented.
FAQ
What is ZCode used for?
ZCode is presented as a GLM-5.2 officially adapted development tool that combines AI agents with an existing toolchain. The public pages show it supporting planning, coding, review, and release in a familiar workflow.
What platforms does ZCode support?
The source pages show ZCode working from a development workspace and completing tasks directly in files, but they do not list a full supported IDE matrix or installation requirements. The changelog does mention desktop-specific fixes for Windows, macOS, and Linux, which suggests the product is available across those platforms.
Does ZCode support plugins or external tools?
The site describes support for skills, plugins, MCP, and AGENTS.md configuration via a built-in configuration guide. The changelog also mentions plugin management, custom plugin addition beta, and a trusted default for workspace MCP servers.
Is pricing available on the public site?
The pricing page at /pricing currently returns a 404, so the public sources here do not confirm a pricing model. Based on the available pages, pricing details should be treated as unavailable rather than assumed.
Can ZCode preserve work when switching tasks?
The source shows ZCode keeping task context such as drafts and attachments when switching tasks, and it supports session recovery and timeline compression improvements in the changelog. Beyond that, the public pages do not give a full account of collaboration features or team workflows.
Quick Facts
- Category
- Developer Tool
- Primary use
- AI-assisted planning, coding, review, and release
- Product type
- GLM-5.2-adapted development tool
- Source domain
- zcode.z.ai
- Pricing page
- /pricing returns 404
- Configurable components
- Skills, plugins, MCP, AGENTS.md
ZCode Alternatives
SonOf
SonOf connects to your repo and PM tool, audits your codebase, and turns approved work into shipped tickets with senior engineering review.
ByteAsk
ByteAsk is a terminal-first AI coding agent for C and C++ that edits repos and verifies changes with compilers, debuggers, sanitizers, and tests.
CreateOS Sandbox
CreateOS Sandbox is an isolated compute environment for running code and agent workloads in Firecracker micro-VMs with private networking and SDK, CLI, or MCP control.
hob
hob is an independent workspace for coding agents, with local control over sessions, terminals, history, routing, and follow-up work.
Manta AI
Manta AI is an autonomous web app testing tool that maps app behavior, catches regressions, and generates tests from a URL, no scripts or selectors needed.
Ghost
Ghost is a terminal-based AI assistant for chatting, code generation, and CLI tasks. Includes free models, supports Linux, macOS, Windows, and is open source.