dcc-mcp-zbrush

Agent workflow

AI agents should use the shared gateway through dcc-mcp-cli; IDE users may
continue to use the MCP endpoint. Prefer typed skills and tools over raw scripts.

Install or update the CLI

dcc-mcp-cli is the preferred control path for every shell-capable agent. If
it is missing, ask the user before installing the latest official release:

# Linux/macOS
curl -fsSL https://raw.githubusercontent.com/dcc-mcp/dcc-mcp-core/main/scripts/install-cli.sh | sh

# Windows PowerShell
powershell -ExecutionPolicy Bypass -c "irm https://raw.githubusercontent.com/dcc-mcp/dcc-mcp-core/main/scripts/install-cli.ps1 | iex"

Keep an official build current through the release manifest:

dcc-mcp-cli update check
dcc-mcp-cli update apply

update apply downloads and stages the latest CLI for the next launch. It
does not update a running dcc-mcp-server; update that server in its own
environment.

dcc-mcp-cli dcc-types
dcc-mcp-cli list
dcc-mcp-cli search --query "<task>" --dcc-type zbrush
dcc-mcp-cli describe <tool-slug>
dcc-mcp-cli call <tool-slug> --json '{"key":"value"}'

dcc-types reports release-catalog support; list reports live sessions. If a
tool belongs to an inactive progressive skill, call dcc-mcp-cli load-skill <skill-name> --dcc-type zbrush before retrying. For post-task improvement,
attach a stable session id with --meta-json, query dcc-mcp-cli stats --range 24h --session-id <task-id>, then pass the bounded evidence to the
review_skill_improvement prompt from dcc-mcp-skills-creator.

ZBrush adapter for the DCC Model Context Protocol ecosystem.

Requires ZBrush 2026.1+ with the official Python SDK (CPython 3.11 embedded in ZBrush).

Quick install

1. Install the Python package

pip install dcc-mcp-zbrush

This installs the dcc-mcp-zbrush Python package — the MCP HTTP server that bridges AI agents to ZBrush. It does not include ZBrush plugin files (see step 2).

2. Install the ZBrush plugin

The plugin files (auto-start script, socket bridge) are distributed separately. Download the plugin ZIP from the latest GitHub Release:

dcc-mcp-zbrush-plugin-<version>.zip

Then run the install script inside the ZIP:

Windows (PowerShell):

.\install\install-windows.ps1

macOS (Terminal):

chmod +x install/install-macos.sh && ./install/install-macos.sh

The installer resolves the ZBrush Asset Directory and installs the recommended
sidecar bridge as Python/init.py. It fails instead of reporting success when
no valid Asset Directory is available; pass -Target explicitly in that case.

3. Restart ZBrush

Launch or restart ZBrush, then start the external MCP sidecar:

dcc-mcp-zbrush --mode sidecar --socket-port 9876

The plugin registers a top-level DCC MCP palette through the official
ZBrush Python SDK with these actions:

• Copy Instance ID
• Server Info
• About DCC MCP

Instance identity belongs to the MCP server. In embedded mode the first two
actions read the public DccServerBase runtime context. The standalone
sidecar plugin does not guess identity from environment variables or registry
files; it directs you to dcc-mcp-cli list when that external context is not
available in the ZBrush process.

4. Health check

Verify the dynamically allocated instance URL:

dcc-mcp-cli list

5. Configure your AI client

Add the MCP server to your AI client config (Cursor, Claude Desktop, etc.):

{
  "mcpServers": {
    "zbrush": {
      "url": "http://127.0.0.1:9765/mcp"
    }
  }
}

How it works

ZBrush does not ship a built-in HTTP REST server. The pre-alpha scaffold that assumed Preferences > Network > Enable HTTP Server was incorrect.

The supported integration paths are:

Rust is not loaded inside ZBrush. The dcc-mcp-core wheel (PyO3) runs
in the external sidecar process; importing its extension module into the ZBrush
2026 embedded VM is not a supported runtime path. The ZBrush-facing bridge is
Python only and executes requests serially on the host main thread while
pumping UI updates.

GoZ C++ SDK is for mesh exchange between DCC apps, not general MCP automation — we do not build the primary adapter on GoZ.

Recommended sidecar mode:

AI Agent → Gateway :9765 → OS-assigned MCP instance → ZBrushMcpServer
         → TCP :9876 → mcp_socket_bridge.py (inside ZBrush) → zbrush.commands

Features (v0.2.0)

• DccServerBase adapter with progressive skill loading
• Bundled skills: zbrush-scripting, zbrush-scene, zbrush-subtool, zbrush-interchange
• In-process executor for ZBrush's embedded Python VM
• Optional socket bridge plugin for sidecar deployments
• Top-level DCC MCP palette registered through zbrush.commands
• Gateway election compatible with dcc-mcp-core

Requirements

• ZBrush 2026.1+
• Python 3.9+ on the sidecar host (ZBrush itself ships 3.11)
• dcc-mcp-core >= 0.19.45

Environment variables

Bundled skills

Path concepts

• PYTHONPATH — where Python looks for packages (pip install handles this)
• ZBRUSH_USER_ASSETS_DIR / ZBRUSH_PLUGIN_PATH — plugin scan roots used by ZBrush 2026.1+

pip install dcc-mcp-zbrush puts the Python package on PYTHONPATH.
The plugin ZIP goes into ZBRUSH_PLUGIN_PATH (handled by the install scripts above).

Skill authoring

Skills lazy-import zbrush.commands and run on the main thread (affinity: main).

from dcc_mcp_core.skill import skill_entry
from dcc_mcp_zbrush.api import import_zbc, with_zbrush, zb_success

@skill_entry
@with_zbrush
def my_tool(**kwargs) -> dict:
    zbc = import_zbc()
    count = zbc.get_subtool_count()
    return zb_success(f"{count} subtool(s)", count=count)

Sidecar mode

1. The plugin ZIP includes sidecar/mcp_socket_bridge.py; install it as
   <Asset Directory>/Python/init.py (install-windows.ps1 -Mode sidecar).
2. Start ZBrush.
3. Run the MCP server outside ZBrush:

dcc-mcp-zbrush --mode sidecar --socket-port 9876

Development

See docs/development.md for source-based setup, testing, and contribution workflow.

References

• ZBrush Python SDK 2026.1
• ZBrush Python environment
• GoZ SDK (mesh exchange only)
• Community reference: newsbubbles/zbrush-mcp (socket bridge pattern)

License

MIT