Annex — ZCode (Z.ai)

This guide covers integrating KaKeKiKoKu with ZCode, Z.ai's desktop agentic development environment. ZCode is designed around long-running software work and is closely integrated with the GLM model family.

The current flagship path is GLM-5.2, which Z.ai presents with a one-million-token context window and multiple reasoning-effort levels. KaKeKiKoKu connects to the ZCode client through MCP; it does not call the Z.ai API or consume model credits by itself.

This guide covers Windows and Linux. ZCode's Linux build is currently distributed through its beta group. Neither combination has yet been validated by the KaKeKiKoKu author on a recorded ZCode and OS version, so the compatibility matrix marks them as Not tested.

ZCode does not support KaKeKiKoKu channels today. KaKeKiKoKu therefore works in Precise mode: GLM can use KaKeKiKoKu MCP tools, while KaKeKiKoKu UI actions that would push messages to the AI copy text to the clipboard for you to paste.

Requirements

Before you start the KaKeKiKoKu integration flow, you need:

• KaKeKiKoKu installed — Windows and Linux instructions are on the Download page. Linux installs from the Snap Store.
• ZCode installed — use Z.ai's official ZCode installation page.
• A model connection in ZCode — ZCode currently offers sign-in paths for Z.ai, BigModel, and API keys. KaKeKiKoKu works through the model connection you choose.

Install or update ZCode

Open the official installation page and choose your platform:

https://zcode.z.ai/en/docs/install

• Windows: download the official x64 or ARM64 installer and complete the setup wizard.
• Linux: join the Linux beta group linked from ZCode's download area and install the package supplied there.

On first launch, choose a project directory as the ZCode workspace and connect your Z.ai account or model provider.

If ZCode is new to you. Ask the Agent to list the files in a small workspace before adding KaKeKiKoKu. This confirms the desktop app and model connection work independently of MCP.

1. Register KaKeKiKoKu as an MCP server

ZCode manages MCP servers in its settings. You can add KaKeKiKoKu directly, import it from OpenCode, or paste a standard JSON configuration.

Option A: Add KaKeKiKoKu in Settings (recommended)

1. Open Settings > MCP Servers.
2. Click New MCP Server.
3. Choose the scope:
   • Workspace is recommended when KaKeKiKoKu should be available only in the current project.
   • User makes KaKeKiKoKu available in every ZCode workspace for that OS account.
4. Enter wuniq as the server name.
5. Keep the type as stdio.
6. Enter the command for your operating system and leave arguments empty:

wuniq-mcp.exe

/snap/bin/wuniq.wuniq-mcp

7. Click Add.
8. Confirm that KaKeKiKoKu is Enabled in the MCP server list.

Windows path note. Use wuniq-mcp.exe. Do not browse to an absolute path under C:\Program Files\WindowsApps\...; Windows protects that folder and external MCP clients can hit Permission denied or EPERM.

Option B: Import KaKeKiKoKu from OpenCode

If you already followed the OpenCode annex and placed KaKeKiKoKu in OpenCode's global configuration, ZCode can import it:

1. Open Settings > MCP Servers.
2. Click the Import icon in the upper-right corner.
3. Select OpenCode as the source. ZCode scans ~/.config/opencode/opencode.json.
4. Choose the destination scope: User or Workspace.
5. Select wuniq and click Import.
6. Confirm that the imported server is enabled.

ZCode leaves the original OpenCode configuration untouched. If KaKeKiKoKu exists only in a project-level opencode.json, add it manually in ZCode instead; the documented importer scans OpenCode's global configuration.

Option C: Full configuration mode

ZCode accepts standard mcpServers JSON in its Full configuration mode:

{
  "mcpServers": {
    "wuniq": {
      "command": "wuniq-mcp.exe",
      "args": []
    }
  }
}

{
  "mcpServers": {
    "wuniq": {
      "command": "/snap/bin/wuniq.wuniq-mcp",
      "args": []
    }
  }
}

For the current UI and import behavior, see ZCode's official MCP Servers documentation.

2. Startup instructions (the second piece of the integration)

MCP registration gives ZCode the KaKeKiKoKu tools. The project instructions file tells ZCode Agent to use them when a new task starts.

1. In the root of the workspace currently open in ZCode, create AGENTS.md if it does not already exist.
2. Keep any existing instructions and add this block:

   # KaKeKiKoKu Studio (Knowledge Engine)
   
   At the start of every session, run `open list` using the KaKeKiKoKu MCP tool. If it returns projects, `open` each one. If the list is empty, proceed normally — the user may open projects later from the KaKeKiKoKu UI. Never assume the working directory is a KaKeKiKoKu project; the working directory may be a container folder holding multiple projects. Only open paths that `open list` explicitly returns.
   
   Before working on any code file, `read` its KaKeKiKoKu sidecar for context. After changing code, run `sync`.

3. Save the file and start a new ZCode task.

ZCode also reads global user instructions from ~/.zcode/AGENTS.md. When both files exist, it loads the global instructions first and the current workspace instructions afterward.

Workspace root matters. ZCode does not merge multiple AGENTS.md files across nested folders and does not continuously read CLAUDE.md. Put the KaKeKiKoKu instructions in the root of the workspace ZCode actually opened.

3. Apply the changes

• MCP settings: after adding or importing KaKeKiKoKu, confirm it is enabled. If it does not appear in the Agent, close and reopen ZCode.
• AGENTS.md: start a new task. ZCode reads workspace instructions when a task begins.
• Workspace scope: make sure the task belongs to the same workspace where you added KaKeKiKoKu and placed AGENTS.md.

If either piece still seems stale, close ZCode completely, reopen the workspace, and begin a fresh task.

4. Start a task and verify

1. Start a new task in the workspace that contains AGENTS.md.
2. If the startup instructions are loaded, ZCode Agent should call KaKeKiKoKu automatically. If it does not, ask:

   Run open list with the KaKeKiKoKu tool.

3. If everything is connected, ZCode will show a KaKeKiKoKu tool call and list the projects KaKeKiKoKu knows about.
4. Let the startup instruction open every returned project, or ask the Agent to open the one you want.
5. In the KaKeKiKoKu UI, the AI-connected indicator should turn green when ZCode uses the tool.

Quick test

Read the project index with KaKeKiKoKu.

GLM should use the KaKeKiKoKu MCP instead of answering from general model knowledge. If it does not, ask explicitly to use the KaKeKiKoKu tool and verify the instruction file location.

5. Working in Precise mode

In Precise mode, KaKeKiKoKu cannot push messages into ZCode. The daily flow is:

• Starting a task. The workspace AGENTS.md should make ZCode Agent open KaKeKiKoKu projects automatically. If not, ask it to run open list with the KaKeKiKoKu tool.
• Sync. When code changes, KaKeKiKoKu detects drift in its panel. Ask ZCode Agent to run sync when you want it to reconcile code and context.
• "Talk to AI" buttons. In Precise mode, these copy a prompt to the clipboard. Paste it into the ZCode task when you want GLM to act on it.
• Drift alerts. They appear in KaKeKiKoKu's UI, not as push notifications in ZCode.

Recommended KaKeKiKoKu configuration. Go to KaKeKiKoKu > Settings > Application > AI Sync and choose Precise mode.

6. Plan and model

ZCode is designed for the GLM family. At the time this page was reviewed, Z.ai positioned GLM-5.2 as its flagship long-horizon coding model with a one-million-token context window. ZCode also supports model connections through Z.ai, BigModel, and API keys.

Model availability, context limits, reasoning modes, prices, and quotas belong to Z.ai and can change without a KaKeKiKoKu release. Check the official ZCode documentation and GLM-5.2 announcement for the current state.

KaKeKiKoKu itself remains model-agnostic. Its search and read tools load project knowledge progressively, so it does not require the whole repository to fit into one prompt.

7. Troubleshooting

KaKeKiKoKu does not appear in ZCode's MCP list

1. Open Settings > MCP Servers and check the correct scope.
2. Verify the server name, type stdio, and OS-specific command.
3. Confirm KaKeKiKoKu is enabled.
4. If you pasted JSON, validate it and make sure there are no duplicate mcpServers keys.
5. Close ZCode completely, reopen it, and open the same workspace.

ZCode cannot import KaKeKiKoKu from OpenCode

The documented importer scans OpenCode's global configuration at ~/.config/opencode/opencode.json. A KaKeKiKoKu entry that exists only in a project's opencode.json may not appear. Add KaKeKiKoKu manually in ZCode or move the entry to OpenCode's global configuration if that matches your preferred scope.

ZCode ignores AGENTS.md

1. Make sure the file is in the root of the current ZCode workspace.
2. Do not rely on a nested AGENTS.md; ZCode does not merge the directory hierarchy.
3. Start a new task after saving the file.
4. Ask explicitly: "Use the KaKeKiKoKu MCP tool to run open list."

"Permission denied", "Access denied" or "EPERM" on Windows

Usually the MCP command points into C:\Program Files\WindowsApps\.... Edit the server and use only:

wuniq-mcp.exe

where.exe wuniq-mcp.exe returns nothing

Windows cannot find the KaKeKiKoKu alias. Check that KaKeKiKoKu is installed, that it opens from the Start menu, and that %LOCALAPPDATA%\Microsoft\WindowsApps is in your PATH. Close and reopen ZCode after installing KaKeKiKoKu.

Linux command is not found

Check that the KaKeKiKoKu Snap is installed and that this command exists:

/snap/bin/wuniq.wuniq-mcp

ZCode and KaKeKiKoKu must run under the same Linux user account. Remember that ZCode's Linux distribution is currently beta.

KaKeKiKoKu cannot see projects outside your home folder on Linux

If projects live under /mnt/, /media/, a USB drive, or a VM shared folder such as /mnt/hgfs/, grant the KaKeKiKoKu Snap access:

sudo snap connect wuniq:removable-media

Then reload the KaKeKiKoKu web UI and open the project again.

ZCode sees KaKeKiKoKu, but KaKeKiKoKu does not show as connected

1. Make sure the task is active and the server is enabled.
2. Ask: "Run open list with the KaKeKiKoKu tool."
3. Watch the KaKeKiKoKu UI. The AI indicator changes only when ZCode actually invokes the MCP.

The model or account fails, but the MCP entry looks correct

Z.ai authentication, plan limits, service availability, and GLM behavior are separate from the local MCP connection. Confirm ZCode can complete an ordinary task, then test KaKeKiKoKu with an explicit open list request.

Still stuck?

Email support@kakekikoku.com with your operating system, ZCode version, MCP scope, configured command, and any visible error. KaKeKiKoKu can also generate a diagnostic report from the Support function in the app.