khz
||𝝞|𝝞||𝝞|𝝞|𝝞||
GitHub - Noizefield/audio-plugin-coder: Audio Plugin Coder (APC) is a groundbreaking, open-source framework that enables musicians, producers, sound designers, and developers to create professional VST3/AU audio plugins using natural language and AI
Audio Plugin Coder (APC) is a groundbreaking, open-source framework that enables musicians, producers, sound designers, and developers to create professional VST3/AU audio plugins using natural lan...
github.com
About Audio Plugin Coder
Audio Plugin Coder (APC) is the result of a long-standing personal obsession: building creative tools, writing music, and ultimately creating professional audio plugins.
While developing software instruments and effects has always been a dream, building real-world VSTs (with robust DSP, UI, state handling, and packaging) is notoriously complex. Over time, and especially with the rapid advancement of AI-assisted development, that barrier has finally crumbled.
Over the past 18 months, APC has been continuously designed, tested, and re-iterated as a practical AI-first framework for building audio plugins. This involved thousands of hours of experimentation, trial-and-error, and yes... occasionally yelling at LLMs to finally render the UI correctly.
Midway through development, I stumbled upon the excellent work of TÂCHES (glittercowboy). His approach to context engineering was a revelation. I adopted some of his core ideas, particularly regarding meta prompting and structured agent workflows and integrated them directly into APC's DNA to create a more robust system.
APC is designed to be Agent Agnostic. Whether you use Google's Antigravity, Kilo, Claude Code, or Cursor, APC provides the structure they need to succeed.
Audio Plugin Coder (APC) is currently in active development. APIs may change, features may be incomplete, and bugs should be expected.
Development Status Disclaimer
Use APC for development and experimentation purposes only until a stable release is announced.
What is Audio Plugin Coder?
Audio Plugin Coder (APC) is a structured, AI-driven workflow system that guides LLM agents through the entire audio plugin development lifecycle.
It enables the creation of VST3 / AU / CLAP plugins using natural language, predefined workflows, and domain-specific skills- without constantly re-explaining context, architecture, or best practices to the AI.
Instead of manually juggling DSP architecture, UI frameworks, build systems, state tracking, and packaging, APC provides a unified framework where AI agents can operate with long-term context, validation, and self-improving knowledge.
Key Features
LLM-Driven Development - Designed to work with Antigravity, Kilo, Claude Code, Cursor, or any coding agent.
Structured Workflows - Five-phase system: Dream → Plan → Design → Implement → Ship.
Dual UI Frameworks - Choose Visage (pure C++) or WebView (HTML5 Canvas).
State Management - Automatic progress tracking, validation, and rollback capabilities.
Self-Improving - Auto-capture troubleshooting knowledge; the system gets smarter over time.
Production Ready - JUCE 8 integration with CMake build system.
Comprehensive Skills - Pre-built domain knowledge for DSP, UI design, testing, and packaging.
Quick Start
Prerequisites
- Windows 11 or Linux (tested with Mint Linux) (macOS not yet tested)
- PowerShell 7+
- Visual Studio 2022 (with C++ development tools)
- CMake 3.22+
- Git
- An LLM coding agent (Claude Code, Antigravity, Kilo, Codex, Cursor)
Vibe-Coding Audio Plugins: From Prompt to VST3 in Minutes (JUCE 8 & APC)
Zuletzt bearbeitet:
