X32/M32 OSC Expert Skill

A source-grounded Agent Skill for Midas M32 and Behringer X32 digital consoles

Exact OSC lookup · Console workflows · Effects · MIDI · Diagnostics · Offline packet tooling · Risk-gated planning

Overview · Capabilities · Install · Examples · Architecture · Validation · Security

---

Overview

x32-m32-osc-expert is a standalone, progressively loaded Agent Skill for technical work involving Midas M32 and Behringer X32 digital mixing consoles.

It combines console-operation knowledge with an exact, source-traceable OSC reference layer. The Skill can explain hardware and control-surface workflows, resolve documented OSC addresses and parameter types, inspect effects and MIDI mappings, encode or decode OSC packets offline, diagnose communication faults, and prepare risk-classified dry-run change plans.

The repository contains the Skill and its offline validation assets only. It does not contain an MCP server, a live console transport, a hosted API, a schema registry, or an autonomous execution service.

Core rule: Never invent undocumented commands. Never derive OSC addresses from UI labels, screenshots, menu paths, MIDI mappings, signal-flow labels, or common conventions.

Why this project exists

X32/M32 integrations fail most often when three different concerns are mixed together:

1. Console operation: what the hardware and UI actually do.
2. Protocol knowledge: the exact OSC address, type, range, direction, and firmware context.
3. Execution authority: whether a state-changing action is permitted, verified, and reversible.

This Skill separates those concerns. It gives an agent a dependable knowledge and validation layer without granting it network access to a console.

Capabilities

Area	What the Skill provides
Console hardware	Control surface, channel strips, banks, layers, faders, encoders, rear-panel I/O, monitoring, talkback, USB, AES50, Ultranet, MIDI and power references
Main display and UI	Screen navigation, editable fields, meters, routing pages, setup, libraries, effects, utility functions and contextual behaviour
Signal flow	Inputs, buses, matrices, mains, inserts, effects, tap points, routing, monitoring and documented performance limits
OSC protocol	Exact address and node lookup, parameter types, argument order, ranges, units, read/write direction, replies and subscriptions
OSC codec	Offline encoding and decoding for documented int32, float32, string and blob forms
Effects	Effect inventory, slot restrictions, parameter tables, enums and documented control ranges
MIDI	MIDI operation, assignments, DAW-control notes and OSC-over-MIDI SysEx references
Diagnostics	Layered troubleshooting for identity, firmware, UDP behaviour, padding, type tags, subscriptions, client contention and state divergence
Safety policy	Read-before-write discipline, four-tier risk classification, dry-run planning and fail-closed handling
Traceability	Source file, page, section, classification, confidence, conflict and parse notes where available

Source authority

The Skill is grounded in two source families:

• The M32 manufacturer user manual for console hardware, UI, routing, effects, MIDI, specifications and block-diagram information.
• The unofficial X32/M32 OSC Remote Protocol reference for firmware 4.0 and later.

The OSC reference is operationally useful, but it is not vendor-authored and explicitly acknowledges possible inaccuracies. The Skill therefore preserves EXPLICIT, DERIVED, UNCERTAIN, CONFLICT, NOT FOUND and UNREADABLE classifications rather than silently repairing or completing the source.

See Data provenance and Licence scope.

Coverage snapshot

Dataset	Included
OSC catalogue rows	753
Explicit OSC rows	724
Quarantined uncertain rows	29
X32node paths	337
Effect inventory rows	95
Unique effect codes	61
Effect parameter rows	705
Unified data-dictionary rows	1,906
MIDI rows	16
Console signal-flow edges	24
Engineering risk entries	10
Documentation gaps	12

Machine-readable catalogues are under skill/x32-m32-osc-expert/data/.

Supported console families

The protocol reference describes the X32/M32 family across Standard, Compact, Rack, Producer and Core variants. Coverage is not identical for every model.

Family or model	Coverage position
Midas M32 Standard	Strongest hardware, UI and operating coverage from the manufacturer manual
Midas M32 Compact / M32C	Protocol-family coverage; verify model-specific surface and I/O differences
Midas M32 Rack / M32R	Protocol-family coverage; no full physical-surface equivalence implied
Behringer X32 Standard	Protocol-family and surface-control reference coverage
Behringer X32 Compact / X32C	Protocol-family coverage with model-specific verification required
Behringer X32 Producer / X32P	Protocol-family coverage with model-specific verification required
Behringer X32 Rack / X32RACK	Protocol-family coverage with model-specific verification required
Behringer X32 Core / X32CORE	Historical protocol coverage; the source notes that the model was discontinued

Always establish the console model and firmware before preparing a state-changing plan.

Skill architecture

flowchart LR
    U[User request] --> A[AI agent]
    A --> S[SKILL.md control plane]

    S --> R1[Hardware and UI references]
    S --> R2[OSC and node catalogues]
    S --> R3[Effects, MIDI and scaling references]
    S --> P[Risk and safety policy]

    R1 --> O[Source-grounded answer]
    R2 --> O
    R3 --> O
    P --> O

    S --> T[Offline deterministic scripts]
    T --> L[Lookup]
    T --> C[Encode or decode]
    T --> V[Validate dry-run plan]

    L --> O
    C --> O
    V --> O

    O --> D[Explanation, diagnosis or dry-run plan]
    D -. no live transport included .-> X[External authorised executor]

Loading

The dotted boundary is deliberate. This repository does not send UDP packets, alter console state, or claim that a transmitted packet was applied.

Repository layout

.
├── README.md
├── LICENSE
├── LICENSE_SCOPE.md
├── SECURITY.md
├── SCOPE_BOUNDARY.md
├── NOTICE.md
├── BUILD_REPORT.md
├── CHANGELOG.md
├── docs/
│   ├── installation.md
│   ├── usage.md
│   ├── publishing.md
│   ├── data-provenance.md
│   └── scope-and-integration.md
├── scripts/
│   ├── verify_repository.py
│   ├── build_release.py
│   ├── publish_github.sh
│   └── publish_github.ps1
├── skill/
│   └── x32-m32-osc-expert/
│       ├── SKILL.md
│       ├── agents/
│       ├── references/
│       ├── data/
│       ├── scripts/
│       ├── tests/
│       ├── evals/
│       └── adapters/
└── dist/
    ├── skill.zip
    └── SHA256SUMS

Installation

ChatGPT

Download dist/skill.zip, then upload it through the Skills interface.

Claude or Claude Code

Copy the canonical Skill directory:

mkdir -p .claude/skills
cp -R skill/x32-m32-osc-expert .claude/skills/

GitHub Copilot or VS Code

Copy the Skill to a supported project-level skills directory:

mkdir -p .github/skills
cp -R skill/x32-m32-osc-expert .github/skills/

Codex and Antigravity

Use the platform-specific instructions under:

skill/x32-m32-osc-expert/adapters/

See the complete Installation guide.

Example requests

Exact OSC lookup

Find the documented OSC schema for channel 1 fader.
Return the address as printed, type tag, range, unit, direction,
firmware scope, source status and evidence. Do not infer missing fields.

Console workflow

Explain how Sends on Fader changes the M32 control surface.
Separate fixed controls, context-sensitive behaviour and display feedback.

Diagnosis

Diagnose why /xremote updates stop after several seconds.
Begin with read-only checks and distinguish timeout, client-limit,
packet-loss and state-reconciliation causes.

Packet review

Decode this OSC packet and compare it with the documented X32/M32 type rules.
Flag alignment, null-padding, endianness and type-tag problems.

Safe change planning

Prepare a dry-run plan to change a documented parameter.
Read the current state first, classify risk, show exact bytes,
define expected verification and provide rollback steps.
Do not send anything.

Deterministic tools

Run tools from the canonical Skill directory:

cd skill/x32-m32-osc-expert

Search OSC addresses

python scripts/lookup_osc.py --query "/ch/01/mix/fader" --json

Search X32node paths

python scripts/lookup_node.py --query "/ch/01" --json

Search effects

python scripts/lookup_effect.py --query "Hall Reverb" --json

Encode an OSC message offline

python scripts/encode_osc.py \
  --address "/ch/01/mix/fader" \
  --types f \
  --args 0.5

Decode a packet offline

python scripts/decode_osc.py --hex "2f696e666f0000002c000000"

Validate a proposed write

python scripts/validate_write.py \
  --address "/ch/01/mix/fader" \
  --value 0.5

These scripts do not open a socket or transmit to a console.

Validation

Run the complete repository validation:

python scripts/verify_repository.py

Expected result:

{
  "status": "PASS",
  "errors": []
}

Build a fresh distributable:

python scripts/build_release.py

Verify its checksum:

shasum -a 256 dist/skill.zip
cat dist/SHA256SUMS

The validation workflow covers:

• external-reference audit
• dataset coverage
• Skill structure and frontmatter
• deterministic unit tests
• clean-extraction validation
• scope-boundary enforcement
• absence of public schema identifiers
• absence of bundled MCP or live-transport implementation

Detailed results are recorded in BUILD_REPORT.md.

Security model

This repository is intentionally offline-first.

It does not include:

• a live UDP or OSC sender
• an MCP server
• a cloud API or hosted endpoint
• authentication or secret management
• firmware-update tooling
• autonomous scene recall
• phantom-power execution
• media-format or destructive operations
• a public JSON Schema registry

Any external execution system must independently provide console identity checks, firmware compatibility, current-state reads, operator approval, rate limits, post-write verification, rollback, audit and a physical override.

Read SECURITY.md and SCOPE_BOUNDARY.md before integrating the Skill into a control system.

Four-tier risk policy

Tier	Meaning	Default policy
Tier 0	Observe and read-only operations	No write approval required
Tier 1	Bounded, reversible, allow-listed changes	Dry run, current-state read, limits and verification
Tier 2	High operational impact	Explicit human approval and recovery plan
Tier 3	Critical, destructive or uncertain	Fail closed; no autonomous execution

Unknown, undocumented, UNCERTAIN, CONFLICT or unreadable write operations never qualify as Tier 0 or Tier 1.

Contributing

Contributions are welcome when they preserve source traceability and the no-invention rule.

Before opening a pull request:

python scripts/verify_repository.py

A contribution that adds or changes a technical fact should include:

• the source document and edition
• PDF and printed page where applicable
• the exact source form
• status and confidence
• a conflict or parse note when needed
• updated tests and coverage data

See CONTRIBUTING.md.

Publishing

Publishing scripts require an explicit owner, repository name and visibility. They perform a safe validation pass before any Git mutation.

Dry run:

./scripts/publish_github.sh \
  --owner DXBMARK \
  --repo x32-m32-osc-skill \
  --visibility private \
  --dry-run

Real publishing requires an exact typed confirmation and verifies the existing origin before creating a commit or pushing.

See Publishing guide.

Licence and attribution

The original scripts, tests, repository automation and authored Skill framework are licensed under the MIT License.

Third-party manuals, trademarks, product names, protocol facts and supplied source documents remain under their original rights. Midas and Behringer are trademarks of their respective owners. This project is independent and is not endorsed by, affiliated with or supported by the equipment manufacturers.

See LICENSE_SCOPE.md, NOTICE.md and CITATION.cff.

Support

• Report reproducible problems through GitHub Issues.
• Review known limitations in BUILD_REPORT.md.
• Read SECURITY.md before reporting a security concern.
• General project information: dxbmark.com.

---

Engineered by DXBMARK LLC

Copyright © 2026 DXBMARK LLC. All rights reserved.