ACE-Step-Studio copilot-instructions.md

Repository guidance for ACE-Step 1.5, an open-source AI model that generates music. It documents the Python and machine-learning stack, supported hardware paths, and the main code areas.

In plain words
What is it for?
Use it when modifying music generation, hardware detection, the Gradio interface, the REST API, dependencies, or platform-specific acceleration.
Why use it?
It helps agents avoid breaking support for one type of processor while changing code for another.

Instructions file for GitHub Copilot

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add instructions/timoncool/ace-step-studio/copilot-instructions
Clone the repo
git clone --depth 1 https://github.com/timoncool/ACE-Step-Studio

Made for: GitHub Copilot.

Per session 732 This file is loaded in full into every session.
When invoked 732 The same file — it is already loaded in full.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.00732 $0.00732
Opus 5 $0.00366 $0.00366
Sonnet 5 $0.00146 $0.00146
Haiku 4.5 $0.00073 $0.00073

Measured 3d ago against content hash a394bdb3cca8, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

ACE-Step-Studio copilot-instructions.md scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 3d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

ACE-Step-1.5/.github/copilot-instructions.md · 68 lines

How it starts

The opening of the file, as written. The whole thing — 68 lines — stays where its author put it; the contents beside it link to each section on GitHub.

ACE-Step 1.5 - GitHub Copilot Instructions

Project Overview

ACE-Step 1.5 is an open-source music foundation model combining a Language Model (LM) as a planner with a Diffusion Transformer (DiT) for audio synthesis. It generates commercial-grade music on consumer hardware (< 4GB VRAM).

Tech Stack

  • Python 3.11-3.12 (ROCm on Windows requires 3.12; other platforms use 3.11)
  • PyTorch 2.7+ with CUDA 12.8 (Windows/Linux), MPS (macOS ARM64)
  • Transformers 4.51.0-4.57.x for LLM inference
  • Diffusers for diffusion models
  • Gradio 6.2.0 for web UI
  • FastAPI + Uvicorn for REST API server
  • uv for dependency management
  • MLX (Apple Silicon native acceleration, macOS ARM64)
  • nano-vllm (optimized LLM inference, non-macOS ARM64)

Multi-Platform Support

CRITICAL: Supports CUDA, ROCm, Intel XPU, MPS, MLX, and CPU. When fixing bugs or adding features:

  • DO NOT alter non-target platform paths unless explicitly required
  • Changes to CUDA code should not affect MPS/XPU/CPU paths
  • Use gpu_config.py for hardware detection and configuration

Code Organization

Main Entry Points

  • acestep/acestep_v15_pipeline.py - Gradio UI pipeline
  • acestep/api_server.py - REST API server
  • cli.py - Command-line interface
  • acestep/model_downloader.py - Model downloader

Core Modules

  • acestep/handler.py - Audio generation handler (AceStepHandler)
  • acestep/llm_inference.py - LLM handler for text processing
  • acestep/inference.py - Generation logic and parameters
  • acestep/gpu_config.py - Hardware detection and GPU configuration
  • acestep/audio_utils.py - Audio processing utilities
  • acestep/constants.py - Global constants

UI & Internationalization

  • acestep/gradio_ui/ - Gradio interface components
  • acestep/gradio_ui/i18n.py - i18n system (50+ languages)
  • All user-facing strings must use i18n translation keys

Training

  • acestep/training/ - LoRA training pipeline
  • acestep/dataset/ - Dataset handling

Read the full file on GitHub · 68 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 3d ago First seen · 68 lines · 732 tokens per session scan A a394bdb3cca8

Subscribe to this mod's changes

ACE-Step-Studio copilot-instructions.md is an instructions file published in the GitHub repository timoncool/ACE-Step-Studio (327 stars, last pushed 10d ago), licensed MIT. It adds 732 tokens to every session, about $0.0037 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.