Oh My Muse Code

The missing productivity layer for Meta Muse Code.

Agents: skip this README and read AGENTS.md.

Oh My Muse Code (plugin id oh-my-musecode, CLI omm) adds a role catalog, slash-commands, session hooks, and durable .omm/ state on top of stock Muse.

It is inspired by oh-my-openagent, oh-my-claudecode, oh-my-codex, and oh-my-grok. It is not a fork. Skills, commands, and hooks are original Muse-native text.

Status

Surface	Ready?
Native plugin (19 skills, 19 commands, 8 hooks)	Plugin-ready — validate with Muse 1.0.1-R2006.1
Ralph Stop loop (`decision: block`)	Shipped — `muse plugins hook test` confirms `should_block: true`
`omm setup` / `omm doctor`	Implemented (local CLI, no deps)
`omm hud`	Implemented — text snapshot of `.omm/` (not a live TUI)
`omm team` `ask` `wait` `mission` `wiki` `update` plus ralplan/interview/verify/autopilot/execute	Implemented — file-based `.omm/` state (not tmux / not remote ask)
Slash-commands `/team` `/ask` `/hud` etc.	Plugin-ready (in-session templates, not a live tmux HUD)

There is no live tmux team dashboard, Claude statusline, or remote ask transport in this release. omm team / omm ask are file-based (roster + SKILL.md description router).

Muse 1.0.1 plugin APIs are experimental. You must set MUSE_EXPERIMENTAL_PLUGINS=1.

Install

Requires Muse Code with experimental plugins enabled.

One-liner (local tree):

MUSE_EXPERIMENTAL_PLUGINS=1 muse plugins install /path/to/oh-my-musecode && MUSE_EXPERIMENTAL_PLUGINS=1 muse plugins approve oh-my-musecode

Marketplace (when the git remote is listed):

export MUSE_EXPERIMENTAL_PLUGINS=1
muse plugins marketplace add omm https://github.com/hypery11/oh-my-musecode
muse plugins install oh-my-musecode@omm
muse plugins approve oh-my-musecode

Validate without installing:

MUSE_NO_AUTO_UPDATE=1 MUSE_LOGIN=0 MUSE_EXPERIMENTAL_PLUGINS=1 muse plugins validate . --json

Companion CLI (optional):

npm i -g oh-my-musecode
omm setup
omm doctor

From a git checkout:

node bin/omm.mjs setup
node bin/omm.mjs doctor

Vanilla Muse vs Oh My Muse Code

	Vanilla Muse Code	With oh-my-musecode
Roles	Bring your own prompts	19 bundled skills (architect through git-master)
Slash-commands	Built-ins only	19 workflow commands (`/ralph`, `/team`, `/verify`, ...)
Session state	Chat + transcripts	Durable `.omm/` plans, memory, traces, verify reports
Hooks	You write them	8 hooks: session, prompt keywords, skill/intent gate, Ralph/ulw/boulder/todo stop-chain, subagent log, compact flush, optional session-end webhook
Multi-agent	`subagent_spawn` + worktrees	Same Muse tools, plus team roster/log conventions
Companion CLI	`muse`	`omm setup` / `doctor` / `hud` plus file-based team/ask/wiki and ralplan/interview/verify/autopilot engines
Experimental flag	Needed for plugins	Documented; required on 1.0.1-R2006.1

Features

Skills (19)

architect, planner, executor, explore, analyst, designer, debugger, tracer, critic, code-reviewer, security-reviewer, code-simplifier, test-engineer, qa-tester, verifier, scientist, document-specialist, writer, git-master

Each skill is a Muse role recipe: when to activate, how to use subagent_spawn and .muse/worktrees/, and which .omm/ files to persist. They do not mention Claude Code or Codex APIs.

Commands (19)

Command	Job
`/team`	Multi-skill mission + roster under `.omm/team/`
`/autopilot`	Plan then execute (CLI autopilot.json; Stop after todo)
`/execute`	One plan step or concrete task (CLI marks plan.json + progress.md)
`/ralph`	Work-until-done loop with iteration budget
`/ralplan`	Ralph-oriented planning (CLI writes plan stub + inactive ralph.json)
`/deep-interview`	Requirements interview (CLI writes interview stamp + requirements.md)
`/ask`	Route a question to the best skill (in-session; CLI scores SKILL.md descriptions)
`/verify`	Evidence-gated completion (CLI pending/pass/fail state files)
`/ultragoal`	North-star goal + first milestone
`/handoff`	Next-session brief
`/skillify`	Draft a new skill from a workflow
`/omm-skill`	Explain a bundled skill
`/hud`	Text snapshot of `.omm/` state (not a live TUI)
`/omm-setup`	Print install/approve steps
`/omm-doctor`	Diagnose Muse binary + plugin tree
`/remember`	Durable notes
`/omm-trace`	End-to-end flow trace
`/wiki`	Lightweight `.omm/wiki/` pages
`/debug`	Structured debugging session

Hooks (8 unique scripts)

Event	Id	Behavior
SessionStart	session-start	Optional system message listing `/commands`
UserPromptSubmit	prompt-keywords	If prompt mentions ralph/ralplan/ultrathink/autopilot, write `.omm/mode.json`
PreToolUse	skill-gate	Optional skill-gate + fail-open intent-gate (plan.json) for mutating tools
Stop	stop-chain	Ralph, then ulw/ultrawork, boulder, capped todo nudge, then autopilot
SubagentStart	subagent-start	Append `.omm/team/log.jsonl`
SubagentStop	subagent-stop	Same log
PreCompact	pre-compact	Write `.omm/compact.json` and append `.omm/memory.md`
SessionEnd	session-end	Audit; optional POST if `.omm/notify.json` has an http(s) URL

Each hook is a distinct Python file (argv uniqueness). Shared helpers live in hooks/_omm.py and are not hook sources. Hooks never emit a bare permissionDecision=allow.

State

Workspace state lives under .omm/ (see .omm/README.md). Plugin source is this repo. Muse worktrees: .muse/worktrees/.

Companion CLI

package.json bin omm -> bin/omm.mjs (Node, no dependencies).

omm setup — print Muse install/approve commands with the experimental flag
omm doctor — locate muse, check this tree, run validate if found
omm hud — text snapshot of .omm/ (not a live TUI)
omm team [mission...] — init/list .omm/team/ (not tmux)
omm ask [question...] — route to a bundled skill via SKILL.md description + id (no remote model)
omm wait [seconds] — poll .omm/team/log.jsonl mtime
omm mission [text...] — queue in .omm/mission/queue.json
omm wiki list|show|write — files under .omm/wiki/
omm update — print muse plugins update/approve; optional registry check
omm ralplan interview ultragoal handoff skillify verify autopilot execute remember debug trace — file-based .omm/ engines (slash-commands still interview in-session)
-h / --help and -V

Layout

.muse-plugin/plugin.json
skills/<id>/SKILL.md
commands/<id>.md
hooks/*.py
bin/omm.mjs + bin/lib/
.omm/                 (runtime; mostly gitignored)

繁體中文（摘要）

Oh My Muse Code 是 Meta Muse Code 缺少的生產力層：19 個技能、19 個斜線指令、8 個 hooks，以及工作區狀態目錄 .omm/。靈感來自 oh-my-openagent、oh-my-claudecode、oh-my-codex、oh-my-grok，不是 fork，內容皆為原創。

Muse 1.0.1 的 plugin API 仍是實驗功能，請設定 MUSE_EXPERIMENTAL_PLUGINS=1，再用 muse plugins install / marketplace / approve。

本版沒有即時 tmux 團隊儀表板或遠端 ask 通道。omm hud 是文字快照；omm team/ask/wait/mission/wiki/update 以及 ralplan/interview/verify/autopilot 等是檔案型 CLI。完整中文說明見 README.zh-TW.md。

Contributing

See CONTRIBUTING.md and the Code of Conduct. Security: SECURITY.md. Roadmap: ROADMAP.md. Honest feature matrix vs OMC/OMX/OMG: docs/FEATURE-MATRIX.md.

Navigation Menu

Repository files navigation

Oh My Muse Code

Status

Install

Vanilla Muse vs Oh My Muse Code

Features

Skills (19)

Commands (19)

Hooks (8 unique scripts)

State

Companion CLI

Layout

繁體中文（摘要）

Contributing

License

About

Releases

Packages

Contributors

Languages

Name		Name	Last commit message	Last commit date
.github		.github
.muse-plugin		.muse-plugin
.omm		.omm
bin		bin
commands		commands
docs		docs
hooks		hooks
scripts		scripts
skills		skills
tests		tests
.gitignore		.gitignore
AGENTS.md		AGENTS.md
CHANGELOG.md		CHANGELOG.md
CONTRIBUTING.md		CONTRIBUTING.md
LICENSE		LICENSE
README.md		README.md
README.zh-TW.md		README.zh-TW.md
ROADMAP.md		ROADMAP.md
SECURITY.md		SECURITY.md
SUPPORT.md		SUPPORT.md
VERSION		VERSION
package.json		package.json

Latest commit

History

Folders and files

Repository files navigation

Oh My Muse Code

Status

Install

Vanilla Muse vs Oh My Muse Code

Features

Skills (19)

Commands (19)

Hooks (8 unique scripts)

State

Companion CLI

Layout

繁體中文（摘要）

Contributing

License

About

Topics

Resources

Code of conduct

Contributing

Security policy

Stars

Watchers

Forks

Releases

Packages

Contributors

Languages