Agent-ready lighting control

AI-friendly Professional DMX Lighting Console

Give your AI agent real lighting tools. Unilighter exposes the open lighting project through a local MCP server so compatible AI clients can inspect the rig, patch fixtures, and build editable scenes, animations, dashboards, and node graphs while you stay in control of the result.

MCP is the control surface for agents

Ask in plain language. Get real console objects.

Connect a compatible AI client to the local MCP server in Unilighter Desktop. The client can inspect the open project and call structured tools for projects, fixtures, devices and patching, scenes, animations, node graphs, universes, dashboards, and device control. It can also search the merged fixture catalogs and import a matching Open Fixture Library, QLC+, or DDF profile.

The result is never trapped in a response: it appears in Unilighter as the same editable building blocks you would create by hand.

UNILIGHTER DESKTOP ONLY MCP runs locally through Unilighter Desktop.

Connected AI clientUNILIGHTER DESKTOP ONLY
Patch these 24 fixtures across two universes, apply the stage tags, and flag any address conflicts.
Done in the open projectComplete patch with validation reportDevices, addresses, universes, and reusable tags are visible in the project

Designed to be understood by machines

The complete agent-ready stack

MCP makes actions possible. Documentation supplies precise product knowledge. The Unilighter Skill teaches an agent how to combine both into a safe lighting workflow.

  1. 01
    Local MCP server

    Structured, discoverable tools operate the project open in Unilighter Desktop.

    DO THINGS
  2. 02
    AI-friendly documentation

    Concise concepts, stable terminology, examples, tool references, and machine-readable paths help agents retrieve the right context.

    UNDERSTAND THE CONSOLE
  3. 03
    Official Unilighter Skill

    A ready-made workflow gives compatible agents setup instructions, proven recipes, and review checks for real show preparation.

    Get setup prompt WORK WELL FROM THE FIRST PROMPT
AGENT WORKFLOWONE-PROMPT SETUP

Teach your agent in one prompt

You don't need to manually configure files or write rules from scratch. Simply paste this instruction into Claude Code, Codex, or your coding assistant (Cursor, Antigravity, Windsurf). Your agent will fetch the official SKILL.md specification and configure itself automatically.

View SKILL.md
Prompt for your AI agent
Read and adopt the Unilighter lighting design & MCP skill from https://unilighter.cc/download/SKILL.md
  • Claude Code & Codex: Paste directly into your prompt or terminal session.
  • Cursor, Windsurf & Antigravity: Send in chat or save to project rules.
  • Specification URL: https://unilighter.cc/download/SKILL.md

From blank project to a show you can tune

The work that used to take dozens of clicks becomes one clear brief

Use AI for the repetitive structure and the ambitious first draft. Keep artistic judgment, verification, and live operation where they belong—with the human operator.

01

Turn an equipment list into a patched rig

Give the agent fixture types, quantities, starting addresses, universes, and roles. It can create devices, assign tags, and surface overlaps before load-in.

02

Create a missing fixture from its manual

Provide the mode and channel table. The agent can build a reusable fixture definition, then patch it into the project for review.

03

Describe a look instead of programming every value

Draft complete scenes from visual intent—fixture groups, colors, positions, and intensity—then tune every property in the console.

04

Generate keyframe animation with musical timing

Ask for a four-bar chase, a measured theatre transition, or staggered movement. Receive editable tracks, keys, curves, and easing.

05

Build procedural and audio-reactive logic

Translate a behavioral idea into a node graph with LFOs, audio bands, smoothing, math, limits, fixture tags, and operator controls.

06

Assemble the operator dashboard too

Put scenes, macros, dimmers, color controls, effect depth, and emergency actions into a practical desktop or mobile control surface.

AI accelerates. The operator authorizes.

Powerful automation with an inspectable handoff

An agent works with named fixtures, properties, tags, keyframes, and nodes—not an opaque blob. Review the patch, fixture channel mapping, limits, and generated show logic before enabling physical output.

✓ Structured tool calls✓ Editable project objects✓ Local Desktop connection✓ Human review before live output

Questions that come up in practice

Common questions

Is the AI built into the lighting console?

The MCP workflow connects Unilighter Desktop to a compatible external AI client. The client interprets your request and calls Unilighter tools through the local MCP server; MCP itself is the tool interface, not the language model.

Can an agent control live DMX output without review?

MCP exposes project and device-control tools, but show-critical changes should remain an operator decision. Inspect patches, fixture channel maps, scenes, and generated logic before sending them to a live rig.

What does the Unilighter Skill add?

The official Skill gives a compatible agent the preferred workflow, safety checks, terminology, and practical recipes for using Unilighter MCP tools effectively instead of discovering the console from scratch. Tell your agent to read https://unilighter.cc/download/SKILL.md to configure Cursor, Claude, Antigravity, or any MCP assistant automatically.

Is a browser-based console fast enough for live shows?

Yes. Standard DMX hardware runs at about 40 frames per second, while Unilighter processes control updates internally at 60 frames per second. This leaves enough headroom for cue triggers, audio-reactive control, and live color changes.

Can I use Unilighter at venues with spotty or no internet?

Yes. Once loaded, the console can keep working offline and the project remains stored locally on the device. Cloud synchronization resumes when an internet connection becomes available.

What is the difference between Free and Pro?

Free includes the project-building tools and real DMX hardware testing, but continuous DMX output stops after 30 minutes. Reopen the project to continue testing. Pro removes the session limit and supports uninterrupted production use.