Agent skill

Muse Plugin Creator

by asgeirtj in asgeirtj/system_prompts_leaks

Creates and validates a new native Muse plugin package in the current workspace, limited to five capability families, and leaves installation to you.

CC0-1.0Auto-check passedAgent Workflows

Install Muse Plugin Creator

skills CLI
$ npx skills add asgeirtj/system_prompts_leaks --skill create-plugin -a claude-code

Project install by default; add -g for ~/.claude/skills/.

GitHub CLI
$ gh skill install asgeirtj/system_prompts_leaks create-plugin --agent claude-code

Project scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).

Manual copy
$ git clone --depth 1 https://github.com/asgeirtj/system_prompts_leaks.git skills-src && mkdir -p .claude/skills && cp -r skills-src/Meta/muse-code/skills/create-plugin .claude/skills/create-plugin && rm -rf skills-src

Use ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.

Claude Code skills documentation · loads skills from .claude/skills/

Facts

Skill name
create-plugin
GitHub stars
69k
Token cost
~1.8k tokens
SKILL.md length
869 words
Files
3 (incl. references)
Skills in repo
121
Repo updated
First seen
Licence
CC0-1.0

At a glance

Creates and validates a new native Muse plugin package in the current workspace, limited to five capability families, and leaves installation to you.

  • Works in 5 steps: the portable plugin ID and human display… → the destination relative to the current… → the requested capability families and IDs; → …
  • Creating a new native Muse plugin with commands, hooks or skills
  • SKILL.md covers Boundaries, Optional References, Clarify First and Portable IDs And Paths, plus 4 more sections
  • Instructions only: no scripts, shell commands, URLs or credentials in SKILL.md

What it does

The agent produces the smallest complete plugin that satisfies the request, in one new destination inside the workspace, and validates it with the installed muse CLI. It never modifies, merges into or deletes an existing path, never installs, enables, trusts, runs or fetches the plugin or its dependencies, never updates an existing plugin and never publishes or writes global configuration.

Before using tools it settles the plugin ID and display name, the destination, the capability families and IDs, and every required artifact and command, asking a focused question when something is missing. Supported families are skills, commands, hooks, mcpServers and reminders; tools, agents, output styles, settings and apps are refused, with custom model tools placed behind an mcpServers entry. IDs follow portable naming rules, and two reference files give the native plugin contract and capability examples.

When your agent uses it

  • Creating a new native Muse plugin with commands, hooks or skills
  • Validating a freshly generated plugin package with the muse CLI
  • Checking whether a requested capability is supported by the native plugin format

Example prompts

  • “Create a Muse plugin called release-helper with one command and a reminder.”
  • “Scaffold a Muse plugin that bundles two skills and validate it with the muse CLI.”
  • “Can a Muse plugin add a custom tool directly? Tell me what is supported.”

Requirements

  • The muse CLI installed, for validation

Workflow steps

5 steps, taken from the first numbered list in SKILL.md.

  1. the portable plugin ID and human display name;
  2. the destination relative to the current workspace;
  3. the requested capability families and IDs;
  4. every required artifact and command;
  5. whether the request would require an unsupported family, dependency fetch, or

What it can do on your machine

Read from SKILL.md and the folder at commit dd45aa5. It shows what the files ask for, not the result of running them.

  • Tool permissions

    Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.

    From allowed-tools in the SKILL.md frontmatter.

  • Runs code

    No scripts in the folder and no shell commands in SKILL.md (its code samples are json).

    From the folder's file list and the shell code blocks in SKILL.md.

  • Network

    No URLs in SKILL.md.

    From URLs in SKILL.md, links to its own repository left out.

  • Credentials

    Names no API keys, tokens, secrets or passwords.

    From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

Context cost

Muse Plugin Creator loads about 1.8k tokens when it runs, and up to ~8.9k if it reads all its reference files. Until then it costs about 74 tokens; SKILL.md has 869 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it
~74
When it runs · the whole SKILL.md, loaded when a task matches
~1.8k
With references · SKILL.md plus every file in references/, read only if the agent opens them
~8.9k

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Auto-check passed

The automated check found no risky patterns in SKILL.md.

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.

SKILL.md

The full file from asgeirtj/system_prompts_leaks at commit dd45aa5, republished under its CC0-1.0 licence (© asgeirtj). 869 words, ~1,813 tokens.

Download SKILL.mdSave it as .claude/skills/create-plugin/SKILL.md (or your agent's skills folder). This skill also uses 2 other files; get the full folder from GitHub.
name
create-plugin
description
Create and validate a new native Muse plugin package in the current workspace. Use ONLY when the user explicitly asks to create a Muse plugin or invokes the create-plugin skill. Do NOT use for application/library plugin classes, third-party plugin systems, or ordinary code changes.

Create Plugin

Create one new native Muse plugin in the current workspace. Use this skill only for Muse native plugin package creation, not for implementing plugin classes or plugin features inside another project. Produce the smallest complete plugin that satisfies the request, validate it with the installed muse CLI, and leave installation to the user.

Boundaries

  • Work only in one new destination inside the current workspace.
  • Never modify, merge into, delete, or replace an existing path.
  • Do not install, enable, trust, execute, or fetch the generated plugin or its dependencies.
  • Do not update an existing plugin. Explain that update behavior needs a separate workflow.
  • Do not publish or write global configuration.
  • Ask before writing when the plugin ID, destination, requested capability, or required command is unclear.
  • Reject unsupported capability families instead of inventing a schema.

The supported capability families are exactly skills, commands, hooks, mcpServers, and reminders. Reject tools, agents, outputStyles, settings, and apps. A custom model tool belongs behind an mcpServers entry, not a direct tools capability.

Optional References

This skill is complete without its references. If read_skill returned a physical SKILL.md path and more detail would help, use read_file to inspect these siblings:

  • references/native-plugin-contract.md
  • references/capability-examples.json

Treat them as read-only guidance. Do not fail merely because a reference was not read.

Clarify First

Before tools, determine:

  1. the portable plugin ID and human display name;
  2. the destination relative to the current workspace;
  3. the requested capability families and IDs;
  4. every required artifact and command;
  5. whether the request would require an unsupported family, dependency fetch, or modification of an existing path.

Ask a focused question for missing facts. Refuse before mutation when the request is outside this skill's boundaries.

Portable IDs And Paths

Plugin and capability IDs must:

  • contain 1 through 80 ASCII bytes;
  • begin with a lowercase ASCII letter or digit;
  • use only lowercase ASCII letters, digits, ., _, and - afterward;
  • not have a case-insensitive basename before the first dot equal to CON, PRN, AUX, NUL, COM1 through COM9, or LPT1 through LPT9;
  • not use a product-reserved plugin ID: loop or muse-core.

Artifact paths must be relative UTF-8 paths beneath the plugin root. Manifest paths use /. Reject absolute paths, .., backslashes in manifest values, symlink escapes, and any path whose canonical parent leaves the current workspace.

Required Manifest Base

Start from this native manifest shape and fill only requested capabilities:

json
{
  "schemaVersion": 1,
  "name": "<portable-id>",
  "displayName": "<human name>",
  "version": "0.1.0",
  "description": "<plain description>",
  "compat": {
    "source": "native",
    "manifestDir": ".muse-plugin"
  },
  "capabilities": {
    "skills": [],
    "commands": [],
    "hooks": [],
    "mcpServers": [],
    "reminders": []
  }
}

The manifest lives at .muse-plugin/plugin.json. Keep empty capability arrays unless the installed validator accepts their omission with zero diagnostics.

For each requested capability, create every path referenced by the manifest:

  • skills: {id, path, enabledDefault?} and a UTF-8 SKILL.md with valid frontmatter;
  • commands: {id, path, enabledDefault?} and a UTF-8 Markdown template;
  • hooks: {id, event, command, timeoutMs?, statusMessage?} and any relative command source named by its argv;
  • mcpServers: a stdio {id, transport?, command} or HTTP {id, transport:"http", url} entry and any relative command source named by its argv;
  • reminders: {id, path, tools?, blocking?, decision, defaultPriority?, maxPriority?, maxChildSteps?, maxInstallsPerRun?, reasoningEffort?, context?} and a UTF-8 duty file. Use the executable decision object, including its V1 envelope and deliveryRole, from references/capability-examples.json.
Show full SKILL.md (362 more words)Show less

Reserve The Destination

No artifact write may occur before all reservation steps succeed:

  1. Resolve the current workspace and requested parent to canonical paths.
  2. Require the parent and proposed destination to remain inside the workspace.
  3. Reject a symlinked destination or a parent whose canonical identity escapes.
  4. Confirm the destination does not exist.
  5. Through the managed shell tool advertised by this session, use one platform-native atomic no-replace directory creation operation. Do not use a check-then-overwrite command.
  6. If creation reports occupied or fails, stop. Never delete or reuse the path.
  7. Canonicalize the new directory and its parent again. Require the same expected identity and workspace containment before the first write_file call.

Use write_file and edit_file for UTF-8 artifacts. Do not use shell redirection to bypass file-tool containment. After reservation, mutate only the new directory.

Validate And Correct

Validate every generated skill directory first. Invoke the installed program with the equivalent structured argv. This is the muse skills validate operation; keep its arguments structured:

json
["muse", "skills", "validate", "<skill-directory>", "--json"]

Then run the muse plugins validate operation with structured arguments:

json
["muse", "plugins", "validate", "<plugin-directory>", "--json"]

A validation result is clean only when the process starts, exits zero, stdout is parseable JSON, top-level valid is true, and diagnostics is present and empty. Warnings are not clean. A missing command, malformed JSON, timeout, denial, cancellation, non-zero exit, valid:false, or any diagnostic leaves the draft incomplete.

For each validation layer, make at most three correction rounds after its first failed result. Edit only the new draft and rerun the same layer. The fourth failed result is terminal for that layer. Never continue to whole-plugin validation while a nested skill is invalid.

Report

On success, report:

  • the canonical artifact path;
  • the capability and file inventory;
  • the number and outcome of nested skill validations;
  • the whole-plugin validation outcome;
  • an unexecuted install proposal as structured program and argv values:
json
{
  "program": "muse",
  "argv": ["plugins", "install", "<canonical-plugin-path>"],
  "executed": false
}

Explicitly state that installation was not run.

On failure or denial after reservation, report the draft path, exact failed operation, remaining diagnostics, and checks not completed. Do not claim the draft is created, valid, complete, or ready to install. Cancellation ends the run without a later report; preserve all pre-existing bytes and let runtime cancellation remain authoritative.

© asgeirtj, CC0-1.0. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

SKILL.md and 2 other files (references) in Meta/muse-code/skills/create-plugin of asgeirtj/system_prompts_leaks.

  • SKILL.md
  • references/capability-examples.json
  • references/native-plugin-contract.md

Open the folder on GitHubat commit dd45aa5

Compare with similar skills

Muse Plugin Creator next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.

Muse Plugin Creator compared with similar skills
SkillStarsUsed inTokensAuto-checkLicenceRepo updated
Muse Plugin Creator this skillasgeirtj/system_prompts_leaks69k—~1.8kAutomated safety check: PassCC0-1.0
Mistral Vibe Plugin Creatormistralai/mistral-vibe5.1k—~3.1kAutomated safety check: PassApache-2.0
Codewhale Skill Creatorcodewhale-hq/Codewhale41k—~1kAutomated safety check: PassMIT
Skill Creatorstacklok/toolhive-studio170—~677Automated safety check: PassApache-2.0
Claude Code Skill Developer Guidediet103/claude-code-infrastructure-showcase10k11 repos~3.5kAutomated safety check: PassMIT
MCP Integration for Pluginsanthropics/claude-plugins-official38k11 repos~3.1kAutomated safety check: PassApache-2.0

Similar skills

  • Mistral Vibe Plugin Creator

    mistralai/mistral-vibe

    Official

    Shows how to build a Vibe plugin package in the Agent Plugins 1.0 format, with a plugin.json manifest and optional skills, MCP servers, hooks and other components.

    5.1k GitHub stars~3.1k tokensUpdated today
    Agent WorkflowsAuto-check passed
  • Codewhale Skill Creator

    codewhale-hq/Codewhale

    Guides creating or improving codewhale skills: file layout, discovery paths, writing rules and whether a need fits a skill, an MCP server, hooks or a plugin.

    41k GitHub stars~1k tokensUpdated today
    Agent WorkflowsAuto-check passed
  • Skill Creator

    stacklok/toolhive-studio

    Create new AI agent skills for Claude Code, Codex, and Cursor.

    170 GitHub stars~677 tokensUpdated today
    Agent WorkflowsAuto-check passed
  • Claude Code Skill Developer Guide

    diet103/claude-code-infrastructure-showcase

    A guide to creating and managing Claude Code skills with auto-activation: skill-rules.json triggers, hooks, enforcement levels, YAML frontmatter and progressive disclosure.

    10k GitHub starsUsed in 11 repos~3.5k tokens
    Agent WorkflowsAuto-check passed
  • MCP Integration for Plugins

    anthropics/claude-plugins-official

    Official

    Explains how to bundle Model Context Protocol servers in a Claude Code plugin, covering config files, stdio, SSE, HTTP and WebSocket server types, and authentication.

    38k GitHub starsUsed in 11 repos~3.1k tokens
    Agent WorkflowsAuto-check passed
  • Claude Code Command Development

    anthropics/claude-plugins-official

    Official

    Explains how to write Claude Code slash commands: Markdown files with YAML frontmatter, arguments, file references, bash context and interactive prompts.

    38k GitHub starsUsed in 10 repos~4.8k tokens
    Agent WorkflowsAuto-check passed

More from asgeirtj/system_prompts_leaks

All 121 skills in this repo
  • Fleet Manager for Agent Sessions

    asgeirtj/system_prompts_leaks

    Shows one digest of coding-agent sessions across your connected machines and lets you open, read, steer, approve, stop and close them, over Herdr, tmux or MSP.

    69k GitHub stars~2.5k tokensUpdated today
    Auto-check passed
  • Muse Code Product Doctor

    asgeirtj/system_prompts_leaks

    Diagnoses a Muse Code installation's own failures from binary and session evidence, instead of treating the report as an ordinary repository bug.

    69k GitHub stars~3.5k tokensUpdated today
    Auto-check passed
  • Agents Project Coordinator

    asgeirtj/system_prompts_leaks

    Runs a goal as a project in which the agent coordinates separate agent threads, judging when to split the work, and interviews you first when nothing can be verified.

    69k GitHub stars~2.9k tokensUpdated today
    Auto-check passed
  • Figma Design Inspector

    asgeirtj/system_prompts_leaks

    Inspects Figma designs through the figma CLI and Figma's MCP server to read variants, spacing, tokens and layouts and to extract assets for implementation.

    69k GitHub stars~936 tokensUpdated today
    Auto-check passed
  • Morning Brief Renderer

    asgeirtj/system_prompts_leaks

    Renders a calm, single-page HTML morning brief from your connected calendar, email and chat, or sets it up to run automatically on weekdays.

    69k GitHub stars~4.5k tokensUpdated today
    Auto-check passed
  • Session Start Hook

    asgeirtj/system_prompts_leaks

    Creating and developing startup hooks for Claude Code on the web.

    69k GitHub starsUsed in 1 repo~1.2k tokens
    Auto-check passed

Categories

Questions about Muse Plugin Creator

What does Muse Plugin Creator do?

Creates and validates a new native Muse plugin package in the current workspace, limited to five capability families, and leaves installation to you. The agent produces the smallest complete plugin that satisfies the request, in one new destination inside the workspace, and validates it with the installed muse CLI. It never modifies, merges into or deletes an existing path, never installs, enables, trusts, runs or fetches the plugin or its dependencies, never updates an existing plugin and never publishes or writes global configuration.

When should I use Muse Plugin Creator?

Muse Plugin Creator fits situations like: creating a new native Muse plugin with commands, hooks or skills; validating a freshly generated plugin package with the muse CLI; checking whether a requested capability is supported by the native plugin format.

How do I install Muse Plugin Creator in Claude Code?

Run `npx skills add asgeirtj/system_prompts_leaks --skill create-plugin -a claude-code`. Or copy the skill folder (Meta/muse-code/skills/create-plugin in asgeirtj/system_prompts_leaks) into .claude/skills/create-plugin in your project. Claude Code loads it when a task matches its description.

How do I install Muse Plugin Creator in Codex?

Run `npx skills add asgeirtj/system_prompts_leaks --skill create-plugin -a codex`. Or copy the skill folder (Meta/muse-code/skills/create-plugin in asgeirtj/system_prompts_leaks) into .agents/skills/create-plugin in your project. Codex loads it when a task matches its description.

Can I use Muse Plugin Creator in Cursor, Gemini CLI or GitHub Copilot?

Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add asgeirtj/system_prompts_leaks --skill create-plugin -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/create-plugin, .gemini/skills/create-plugin, .github/skills/create-plugin and .opencode/skills/create-plugin in your project.

What does Muse Plugin Creator need to run?

SKILL.md names no scripts, command-line tools or credentials: Muse Plugin Creator is instructions for the agent only. Our summary lists: The muse CLI installed, for validation.

Does Muse Plugin Creator access the network?

SKILL.md contains no URLs. Any network use would come from the scripts or tools the agent runs. This is read from the text; nothing was executed.

Is Muse Plugin Creator safe to install?

Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. Review the folder before installing.

What licence does Muse Plugin Creator use?

Muse Plugin Creator is published under the CC0-1.0 licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.

How many tokens does Muse Plugin Creator use?

About 1.8k tokens (SKILL.md is roughly 7.3k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full. Its references folder adds about 7.1k tokens, read only when the agent opens those files.

What are the alternatives to Muse Plugin Creator?

Skills that share tags, products or a category with Muse Plugin Creator: Mistral Vibe Plugin Creator (mistralai/mistral-vibe, 5.1k stars), Codewhale Skill Creator (codewhale-hq/Codewhale, 41k stars), Skill Creator (stacklok/toolhive-studio, 170 stars) and Claude Code Skill Developer Guide (diet103/claude-code-infrastructure-showcase, 10k stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.

Who maintains Muse Plugin Creator?

asgeirtj (a GitHub user) maintains it in asgeirtj/system_prompts_leaks, which has 69,095 GitHub stars. The repository holds 121 skills in this directory. The repository was last updated on October 7, 2026.

Source: asgeirtj/system_prompts_leaks on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.