How to use Claude to light MIDI pads for OBS scenes
Claude helped turn MiniLab pads into OBS scene controls, using real MIDI feedback to fix velocity bindings, LED addresses, and silent WebSocket callbacks.
Suggested prompt
Suggested prompt written by Agents In Use, not the source author. Replace bracketed inputs and review tool permissions first.
Connect my Arturia MiniLab 3 to OBS scene switching and active-scene LED feedback. Inspect real MIDI messages, ignore pad velocity where appropriate, edit scene UUID bindings, and verify User mode. Implement SysEx LEDs with the correct pad index offset and OBS WebSocket callback, pausing for my physical button tests.
Reach people exploring AI agents
See placements →Evidence at a glance
Sourced account- Source author
- Сережа Рис
- Original account
- Read the source ↗
- Published
- Source link checked
- Reported result
- The MiniLab pads switched OBS scenes and illuminated the active scene after velocity, LED-index, and callback corrections.
- Human involvement
- The owner specified scene mappings, pressed the physical pads, observed which buttons lit, and reported failures back to Claude.
A linked account supports attribution. It does not mean we independently reproduced its result.
Report an issue with this case
Open correction emailOpens your email app with the case link included. Nothing is sent automatically.
How the agent did it
- Claude researched the OBS MIDI plugin and used MIDI Monitor feedback to diagnose force-dependent pad messages.
- It changed JSON bindings to ignore velocity and wrote a Python SysEx scene indicator.
- The owner tested real pads while Claude corrected LED indices, event callbacks, and documented User mode.
Сережа Рис wanted an Arturia MiniLab 3 to switch OBS scenes during streams and illuminate the pad for the active scene. Claude Code researched an OBS MIDI plugin, inspected configuration, and wrote the Python indicator while the owner tried the controls on the actual keyboard.
The first binding failed because the pads sent different velocity values depending on how firmly they were pressed. MIDI Monitor exposed that behavior, and ignoring velocity in the binding made scene selection work. Claude then added another binding directly in the plugin’s JSON configuration.
The lighting path needed its own corrections. A first attempt illuminated Shift and Hold because pad LED indices began at four rather than zero. Polling was replaced with OBS WebSocket events, and the callback had to use the event-specific name expected by the Python client. The owner also documented that the keyboard must be in User mode.
The reported result was working scene pads and a synchronized active-scene indicator. The account is useful because it connects software changes to physical symptoms: force-sensitive input, incorrect LED addressing, and silent event callbacks. It describes this owner’s macOS and MiniLab setup, rather than proving compatibility with every MIDI controller.
Prompt
Suggested prompt (written by Agents In Use, not the author's)
Source: Personal blog · Сережа Рис
Frequently asked questions
Why did the first MIDI binding fail?
It matched a fixed velocity while actual pad presses varied.
Which LEDs were addressed incorrectly?
Indices starting at zero reached other buttons; pads began at four.
Why was the callback silent?
The Python event client required the event-specific callback name.
More details
About this account
Claude researched the OBS MIDI plugin and used MIDI Monitor feedback to diagnose force-dependent pad messages. It changed JSON bindings to ignore velocity and wrote a Python SysEx scene indicator.
Original Russian first-person post. Actual physical feedback is reported by the author. No original complete task prompt is quoted; the proposed workflow is ours. Suggested prompt written by Agents In Use; no original task prompt was published in the verified account.
Credit: Сережа Рис — Personal blog
Source published
Added to Agents In Use
Source checked
Tools and topics
Claude Code · Arturia MiniLab 3 · OBS · MIDI Monitor · Python
MIDI · hardware debugging · streaming controls
Editorial completeness: 100/100. Measures documented fields, not agent performance or verified outcomes.