English | 中文

Website · Marketplace · Showcase · For Agents · Ecosystem

Connect AI agents to creative tools for scene editing, asset processing, and production workflows.

DCC-MCP helps artists, technical artists, and tool developers connect AI agents to applications such as Maya, Blender, and Houdini. DCC stands for Digital Content Creation; MCP stands for Model Context Protocol, an interface between AI applications and external tools. An agent can call these tools to carry out a task and check the results.

Alongside desktop DCC applications, DCC-MCP connects game engines, 2D design tools, production systems, asset services, profilers, and custom studio software. They share application discovery, tool calls, permission checks, and monitoring. DCC-MCP provides the connections; you choose the agent.

A Skill is a task guide for agents. Start with dcc-mcp to operate an integrated application, or choose a creator Skill below to develop adapters or Skill packages. See the agent guide for installation and configuration.

The dcc-mcp Skill guides agents through dcc-mcp-cli: find a running application and its available tools, read the selected tool's parameter definition (schema), validate the arguments, call the tool, and check the result. These typed tools define their inputs, outputs, and argument types so agents can inspect requirements before calling them. When a call fails, the Skill guides agents to preserve traces and submit a sanitized report to the responsible Skill, adapter, or Core repository.

Find guides and adapters for Maya MCP, 3ds Max MCP, Blender MCP, Unreal MCP, Unity MCP, Tuanjie MCP, and Godot MCP below. Existing Maya CLI, 3ds Max CLI (3dsmax or 3ds max), and Blender CLI workflows can continue to use each application's command line. DCC-MCP adds application discovery, tool search, parameter inspection, calls, and result verification.

• Unreal Engine: inspect and author levels, assets, Blueprints, cinematics, and effects, then validate through PIE or Unreal Automation. On supported editors, the adapter can bridge installed Epic toolsets while Epic continues to own and distribute its experimental MCP plugin.
• Unity: inspect project and scene state, make bounded Undo-backed editor changes, run tests, capture Play Mode, and produce verified builds through the Unity adapter's typed tools.
• Tuanjie / 团结引擎: inspect the optional native Codely CustomTool catalog before using its image, sprite, 3D, material, audio, video, or terrain generators. Tuanjie continues to own sign-in, credits, downloads, and task recovery; DCC-MCP owns the surrounding discovery, routing, and verification.
• Godot: inspect and edit projects and scenes, exercise editor and game runtime tools, and validate 2D or 3D results through the Godot adapter.

A model can write and run a mayapy, hython, or Blender Python script to demonstrate a DCC task. Everyday production also needs permission controls, parameter checks, execution records, and verification of the results.

Generating fresh code for each task means reviewing whether it fits the current scene. Adapter authors also repeat work on transport, main-thread dispatch, process management, instance selection, and logging.

DCC-MCP brings those shared functions into a reusable framework:

Teams can maintain their existing DCC interfaces, permission rules, and production workflows, then reuse those tools when changing agents or models.

DCC-MCP uses MCP to connect agents and can connect creative applications through Python, C++, HTTP, command ports, or native plugins. Stable application APIs can join the same discovery, execution, and management workflow.

Vendor tools can join that workflow too. Unreal Engine 5.8 introduced an experimental official MCP server and Toolset Registry. Our unreal-official-mcp Skill enables and bridges those official toolsets without redistributing Epic's plugin or changing its tool names and schemas. Vendor tools, DCC-MCP tools, and studio tools can share one agent-facing workflow.

Unity projects can use the same pattern with Tuanjie Editor's optional AI packages. The unity-tuanjie-ai Skill inspects the native Codely CustomTool catalog and only executes a tool from a fresh inspection result. Tuanjie's packages continue to own sign-in, credits, downloads, and task recovery; DCC-MCP provides the typed discovery, routing, and verification workflow around those vendor capabilities.

Some older tools have no API and cannot be modified. Other operations are only available through windows, dialogs, or embedded web pages. dcc-mcp-core provides UI Control for these tasks through Qt, native accessibility, webviews, or application-specific UI backends. UI actions use the project's DCC-CUA and UI Control routing. The workflow follows a snapshot -> find -> act -> wait -> verify loop.

Native Skills and APIs remain preferred. UI actions must stay scoped, policy-checked, audited, and verified; whole-desktop access is denied by default. See the UI Control workflow guide.

A Skill can combine task instructions, tool definitions, and existing scripts to make a team's tested production workflows available to agents. Each Skill can be versioned, tested, and distributed independently.

Clear tool descriptions and parameter definitions let agents reuse operations instead of generating code repeatedly for the same task. Parameter validation and result checks remain necessary; outcomes depend on the tool, model, and task. Large studios can distribute different Skills by project, department, or production stage.

Technical directors (TDs) and technical artists (TAs) can package naming rules, scene checks, asset preparation, pre-publish checks, cache/export rules, and review hand-offs in SKILL.md, tools.yaml, and existing studio scripts. Core and the adapter handle connectivity, main-thread execution, routing, permissions, and execution records. Teams can distribute Skills through a public or private Marketplace while reusing the same infrastructure and adapters.

The official Marketplace makes those optional capabilities searchable, installable, upgradeable, and suitable for private studio catalogs too. Browse real outputs and reusable prompts, or inspect the catalog source.

Find the extensions below in the official catalog. Agents should search the latest catalog to confirm package availability before installation:

dcc-mcp-cli marketplace search --query "<capability>" --limit 20
dcc-mcp-cli marketplace install <package_name> --dcc <dcc_name>

The Gateway Admin UI exposes calls, traces, logs, health, statistics, and usage activity. Teams can see which tools agents selected, where calls failed, and which Skills are used most often. Use those records to improve a description, parameter definition, or implementation, then check the changes through real calls.

Browse the DCC-MCP repositories for more adapters, asset services, UI automation, profilers, and production Skills.

Core provides shared connectivity and runtime services. Adapters and Skills implement each application's production rules. If an application has no transaction API, Core cannot guarantee safe rollback.

Try DCC-MCP in your own production workflows. If it helps, support the project with a Star.

Issues and pull requests are welcome. Please report bugs, integration gaps, and production workflow needs in the relevant repository. Focused adapters, Skills, tests, documentation, and interoperability improvements are especially useful.