NoizefieldNoizefield/audio-plugin-coder/AGENTS.md

audio-plugin-coder

AGENTS.md334 starsChanged 34 days agoMITView on GitHub

These instructions apply to the entire repository and are written for any coding agent that reads the AGENTS.md standard (Codex, Cursor, Pi, Antigravity, OpenCode, and others). Canonical APC knowledge lives under .agents/ (skills/, workflows/, rules/, guides/, troubleshooting/). Host-specific folders (.claude/ for Claude Code, .kilocode/ for Kilo, .agent/ legacy, .opencode/command/ for OpenCode, .devin/skills/ for Devin, .pi/ if present) contain pointers to those canonical files; treat them as equivalent.

Where files are headingNew AGENTS.md files have more rules about keeping secrets out of the code, and fewer about code style.
Of the AGENTS.md files we read, the 31,774 first seen in the last six months, against the older ones: the kind of rule that gained most share, and the kind that lost most. About files like this in general, not this one, and an estimate, since a kind is a pattern in the words.
In this repository33 other files it gives its agents.
- .cursor/rules/no-ai-attribution.mdcCursor rule
- apc-setupSkill
- audio-plugin-coderSkill
- debugSkill
- designSkill
- dreamSkill
- implSkill
- orchestrateSkill
- planSkill
- shipSkill
- skill-debugSkill
- skill-designSkill
- juce-webview-windowsSkill
- skill-ideationSkill
- skill-implementationSkill
- skill-packagingSkill
- skill-planningSkill
- skill-testingSkill
- skill-troubleshootingSkill
- testSkill
- apc-setupSkill
- skill_debugSkill
- skill_designSkill
- dreamSkill
- implSkill
- orchestrateSkill
- planSkill
- shipSkill
- juce-webview-windowsSkill
- skill_testingSkill
- skill_troubleshootingSkill
- testSkill
- audio-plugin-coderSkill

The file

Show its rules and text

We split this AGENTS.md into the rules it gives an agent

21 rules. About 0 are common, 8 are distinctive, and 13 are very specific to this project.

≈0
Common · estimate
Many files like this have a rule of this kind.
≈8
Distinctive · estimate
This author's own choices. Worth a look.
13
Very specific
Tied to their project's names, files or tools. Probably skip.

Common
Distinctive
Very specific

Very specific:
Do not add Co-authored-by trailers for AI agents.
line 16
Distinctive:
Do not list AI agents as contributors, authors, or acknowledgments.
line 17
Distinctive:
Do not post GitHub comments, reviews, or issues in the name of an AI agent.
line 18
Distinctive:
If a host injects an AI co-author trailer into a commit message, strip it before the commit is pushed.
line 19
Very specific:
Use the audio-plugin-coder skill for APC lifecycle work.
line 23
Very specific:
If the agent host does not expose slash commands, use the audio-plugin-coder skill or an equivalent natural-language request.
line 28
Very specific:
Never use Codex /plan or /status as APC workflow commands.
line 34
Very specific:
Before changing a plugin under the configured plugins directory, read its status.json.
line 40
Very specific:
Read the relevant workflow and skill under .agents/.
line 42
Very specific:
Use PowerShell and .ps1 scripts on Windows.
line 50
Very specific:
Use Bash/Zsh and .sh scripts on macOS or Linux.
line 51
Distinctive:
Do not copy a PowerShell example from an APC workflow verbatim on macOS or Linux.
line 52
Very specific:
Update status.json through the platform state-management script when a phase completes.
line 59
Distinctive:
Stop after the requested phase instead of automatically starting the next phase.
line 60
Very specific:
If setup.completed is false, warn once and suggest /apc-setup (do not hard-block).
line 61
Distinctive:
Run build operations from the repository root.
line 65
Very specific:
Do not invoke raw cmake, xcodebuild, msbuild, or compiler commands for normal APC builds.
line 66
Very specific:
Use node bin/apc.js build <Name> (preferred, cross-platform) or scripts/build-and-install.ps1 on Windows / scripts/build-and-install.sh on …
line 67
Very specific:
Check the framework version with node bin/apc.js version when reporting issues (canonical source: package.json).
line 68
Distinctive:
Start with the narrowest validation relevant to the changed plugin.
line 69
Distinctive:
Do not alter unrelated generated plugins, build artifacts, or user debug output.
line 70

≈ “Common” is an estimate: at least 1 in 5 of the 40,246 AGENTS.md files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.

Files

AGENTS.md

Audio Plugin Coder — Agent Guidance

Scope

These instructions apply to the entire repository and are written for any coding agent that reads the AGENTS.md standard (Codex, Cursor, Pi, Antigravity, OpenCode, and others). Canonical APC knowledge lives under .agents/ (skills/, workflows/, rules/, guides/, troubleshooting/). Host-specific folders (.claude/ for Claude Code, .kilocode/ for Kilo, .agent/ legacy, .opencode/command/ for OpenCode, .devin/skills/ for Devin, .pi/ if present) contain pointers to those canonical files; treat them as equivalent.

First run

New checkouts should start with /apc-setup (skill action setup) before creating plugins.
Setup writes machine-local apc.config.json (see apc.config.example.json). Paths, UI defaults, and per-phase model preferences live there.
Resolve plugin/build/release directories via scripts/lib/Get-ApcPaths.ps1 or scripts/lib/apc-paths.sh — do not assume ./plugins if config overrides exist.
/apc-ship writes installers and zips to paths.release_dir (default release/). Do not use dist/ as the ship output.

Attribution

Do not add Co-authored-by trailers for AI agents.
Do not list AI agents as contributors, authors, or acknowledgments.
Do not post GitHub comments, reviews, or issues in the name of an AI agent.
If a host injects an AI co-author trailer into a commit message, strip it before the commit is pushed.

APC Workflow

Use the audio-plugin-coder skill for APC lifecycle work.
Primary slash commands are prefixed: /apc-setup, /apc-dream, /apc-plan, /apc-design, /apc-impl, /apc-test, /apc-debug, /apc-ship, /apc-status, /apc-resume, /apc-new, /apc-orchestrate.
Short forms (/dream, /design, /impl, …) remain as deprecated aliases that redirect to /apc-*.
Claude Code and Kilo use the bare /apc-* forms. Codex repo sessions use the audio-plugin-coder skill action; plugin surfaces may expose namespaced command adapters, but they do not create bare /apc-* aliases.
Devin exposes repo skills as slash commands via .devin/skills/apc-*/SKILL.md pointers to .agents/workflows/apc-*.md. /apc-setup is namespaced (/devin:apc-setup, /agents:apc-setup, /claude:apc-setup) because same-named skills exist in multiple imported dirs.
If the agent host does not expose slash commands, use the audio-plugin-coder skill or an equivalent natural-language request.

Codex

Invoke APC as $audio-plugin-coder:audio-plugin-coder <action> <PluginName> or use an equivalent natural-language request.
Actions include setup, dream, plan, design, impl, test, debug, status, resume, ship, new, orchestrate (also accept apc-dream style names by stripping the apc- prefix).
Never use Codex /plan or /status as APC workflow commands; those names are reserved by Codex built-ins. Use the skill actions above.
See docs/codex-compatibility.md for setup and the full command mapping.
Optional cost-aware CLI routing (ChatGPT-authenticated codex exec, Luna→Terra→Sol→Astra): see docs/codex-orchestration.md and scripts/codex/apc-codex-run.ps1 / .sh. Do not force this path on Claude, Cursor, or Kilo sessions.

Required Context

Before changing a plugin under the configured plugins directory, read its status.json.
Resolve the plugin directory with Get-ApcPluginPath / apc_plugin_path (from scripts/lib/Get-ApcPaths.ps1 or scripts/lib/apc-paths.sh). Do not hardcode plugins/<Name> — honor paths.plugins_dir in apc.config.json.
Read the relevant workflow and skill under .agents/; host folders (.claude/, .agent/, .kilocode/) contain pointers to the same canonical files.
Also read .agents/rules/juce-build-protocols.md and .agents/rules/file-naming-conventions.md before implementation, build, or packaging work; read .agents/rules/ui-layout-system.md before any WebView UI layout, scaling, or design-contract work.
Preserve the selected ui_framework: Visage work must not introduce WebView files, and WebView work must not introduce Visage controls.
Announce the preferred model from apc.config.json → models.phases.<phase> at phase start (see docs/model-routing.md).

Platform Rules

Detect the host OS before choosing commands.
Use PowerShell and .ps1 scripts on Windows.
Use Bash/Zsh and .sh scripts on macOS or Linux.
Do not copy a PowerShell example from an APC workflow verbatim on macOS or Linux; use the equivalent shell function or script.

Phase Gates

Complete only the requested APC phase.
Validate the prior phase before writing files.
Back up plugin state before implementation, debugging, or packaging changes.
Update status.json through the platform state-management script when a phase completes.
Stop after the requested phase instead of automatically starting the next phase.
If setup.completed is false, warn once and suggest /apc-setup (do not hard-block).

Build and Validation

Run build operations from the repository root.
Do not invoke raw cmake, xcodebuild, msbuild, or compiler commands for normal APC builds.
Use node bin/apc.js build <Name> (preferred, cross-platform) or scripts/build-and-install.ps1 on Windows / scripts/build-and-install.sh on macOS/Linux (they honor apc.config.json paths).
Check the framework version with node bin/apc.js version when reporting issues (canonical source: package.json).
Start with the narrowest validation relevant to the changed plugin.
Do not alter unrelated generated plugins, build artifacts, or user debug output.
APC targets JUCE 9 (_tools/JUCE). WebView interop: @juce-framework/webview or native/typescript/webview-interop/dist/index.js.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

It worked
Partly
It didn't

What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

Check your setup, free