AI Agent Hub
Back to skills
DCC-MCP Adapter Creator icon

DCC-MCP Adapter Creator

Development Updated 2026.08.30

Paste the following prompt into your AI chat to install this skill:

Please install @user_15292d5a/yjkj-dcc-mcp-creator according to https://skillhub.cn/install/skillhub.md.

About this skill

Problem

When adding MCP adapters to DCC applications, the hard part is not one script; it is coordinating host startup hooks, per-DCC services, the Rust dcc-mcp-sidecar, a machine-wide dcc-mcp-server gateway daemon, readiness, remote registration, and diagnostics into a maintainable path. dcc-mcp-creator is for creating new DCC-MCP adapters or modernizing existing adapter repositories, covering server composition, host-thread dispatch, sidecar/gateway wiring, readiness, resources, project state, diagnostics, install lifecycle, and cross-DCC verification. If the task is only an individual skill package, the docs point to dcc-mcp-skills-creator instead.

How it works

It starts by classifying the host: Embedded Python host such as Blender, Houdini, Maya, and Nuke; External bridge host such as ZBrush, Photoshop, and Unity; and mixed Python/C++ hosts such as Unreal or Unity. Then it recommends starting from DccServerBase and DccServerOptions.from_env(...), routing host API calls through HostExecutionBridge instead of hand-rolling another script executor. Core capabilities include data-driven DCC identity, helper-based skill discovery, minimal mode, resources, diagnostics, context snapshots, install lifecycle, and gateway failover. For gateway run modes, it distinguishes serve, auto, gateway, --no-auto-gateway, and --legacy-gateway-election, with attention to gateway_runtime_mode, guardian behavior, /v1/readyz, and sidecar readiness contracts. When a startup hook must spawn a sidecar, use build_sidecar_command(...) or launch_sidecar(...) to produce the canonical dcc-mcp-server sidecar argv rather than constructing ad-hoc binary commands. For same-LAN or NAT/routed deployments, the docs describe HTTP registration, mDNS discovery hints, and relay/tunnel metadata requirements.

Notes

This skill is closer to an adapter engineering playbook than a general MCP tutorial. Do not treat sidecar spawn success as proof that tools/call is usable; verify dispatch_ready, readiness status, and the host RPC bridge. stub:// is test-only and should not be used as startup proof. Remote registration, mDNS, and relay paths need explicit auth/TLS and caller-attribution handling. It fits teams that already have DCC-MCP core, gateway, and sidecar runtime and want to reduce duplicated adapter glue.

Use Cases

  • When adding a DCC-MCP adapter for Blender, Houdini, or Maya, design startup hooks, per-DCC services, and sidecar wiring.
  • When refactoring an adapter, move host script execution to HostExecutionBridge and add readiness plus gateway failover diagnostics.
  • Before release, verify sidecar dispatch by checking dispatch_ready, /v1/readyz, and the host RPC bridge to avoid false readiness.
  • When deploying DCC-MCP across NAT or routed subnets, configure HTTP registration, same-LAN mDNS discovery, or relay/tunnel metadata.

Best For

  • An integration engineer building Blender, Houdini, or Maya plugins needs a new adapter that can register with the DCC-MCP gateway.
  • A platform engineer maintaining DCC-MCP core or gateway wants to consolidate duplicated sidecar and readiness logic across DCC adapters.
  • A technical owner for Unreal or Unity host bridges needs to connect host RPC to the sidecar and verify dispatch_ready.
  • A release engineer preparing DCC adapter releases needs smoke paths, diagnostics, and release checklist checks.