Create Package logo

Create Package

Community
lploc94
create-package

Use when packaging or sharing Raw agents, tools, skills, hooks, vars or MCP as a .rawpkg, or preparing an updated release. Covers CLI and dashboard exports, recipient inputs and installation checks.

Overview

Publisherlploc94
Repositoryraw-cli
Skill namecreate-package
Stars
107
Forks
14
Bundled files
3
LicenseMIT
Links
  • Markdown instructions

    A SKILL.md file the model loads on demand, so it only costs tokens when a request actually matches.

  • Works with any LLM

    AI skills are plain Markdown, not provider-specific code, so this works with GPT, Claude, Gemini, Grok, or a local model.

  • 3 bundled files

    Scripts, templates, and references the model can read while it works. Files are read-only and never executed.

  • Open source

    Published by lploc94 on GitHub. Read the source before you install it.

Installation

Install the Create Package AI skill in TypingMind to use it with any LLM, or drop it into another agent that reads SKILL.md.

1

Install in TypingMind

TypingMind installs a skill straight from its GitHub folder — it reads SKILL.md, bundles the resource files, and stores the result locally.

  1. Open the app and go to Plugins → Skills.
  2. Choose "Install from GitHub".
  3. Paste the skill folder URL below and confirm.
    https://github.com/lploc94/raw-cli/tree/main/examples/skills/create_package
  4. Enable the skill in any chat where you want it available.
Plugins → Skills → Add skill → From GitHub URL, then paste the folder URL and press Continue.
2

Install in another agent

Any agent that reads the Agent Skills format can use this skill — copy the folder into that agent's skills directory.

Claude Code — .claude/skills
git clone --depth 1 https://github.com/lploc94/raw-cli.git /tmp/raw-cli
mkdir -p .claude/skills
cp -r /tmp/raw-cli/examples/skills/create_package .claude/skills/create-package
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Create Package in any TypingMind chat and the model takes it from there. Its name and description sit in the system prompt, and the moment a request matches, the model loads the full instructions itself — you never invoke it by hand, and it costs no tokens until it is actually used.

The model loads Create Package on its own as soon as a request matches it.

Works with any AI model

AI skills are plain Markdown instructions rather than provider-specific code, so Create Package is not tied to the model it was written for. Install it once in TypingMind and use it with GPT-5, Claude, Gemini, Grok, DeepSeek, Mistral, Llama, or a local model you run yourself — all on your own API keys.

  • Loaded only when it is needed

    The system prompt carries just the name and description. The instructions are fetched on the first matching request, so an idle skill costs nothing.

  • Switch models mid-chat

    Because the skill is instructions rather than code, changing model does not break it — the next model reads the same SKILL.md.

Skill instructions

This is the SKILL.md content the model loads. Read it before installing — a skill is instructions your model will follow.

Create a shareable Raw package

Produce editable package source, a validated .rawpkg archive and instructions the recipient can follow. Use for packaging existing components or assembling a package. Creating a new tool, skill, agent or MCP connection belongs to its creation skill when selected; use that guidance only for components the request actually needs.

For a how-to, explain the relevant commands and contract. For a requested artifact, carry out the workflow below. For a failed package, start from its validation or activation error and preserve unrelated configuration.

For browser import/export and typed binding forms, read references/dashboard.md with read_file.

Choose the source and contents

Establish the source agent or component paths, output directory, scoped package name (@owner/name) and SemVer release (1.0.0). Keep the requested scope: a tool-only or skill-only package needs no agent export. An export name, installed alias and recipient agent name are separate identifiers.

  • Existing agent: raw package export --agent NAME --name @owner/name --version 1.0.0 --out DIR --config AUTHOR_CONFIG. Use a new or empty destination. This copies selected owned assets and produces a report; inspect required inputs, external executables and unresolved assets before packing.
  • Selected components or a new composition: author raw-package.json and its declared files. Start from the installed examples/packages/tool-only/, skill-only/ or mixed-kit/ when helpful. The examples and docs/packages.md belong to the installed raw-cli package; do not assume a source checkout or resolve them from the project cwd.

Before writing a manifest or inputs, read references/manifest.md with read_file. Before installation, activation, updates or development linking, read references/packages.md. Resolve these references beside this loaded skill; load_skill does not automatically include their contents.

Package prompts, helpers, skill resources and provider/MCP scripts that the package owns. Keep recipient model aliases, machine paths, environment names and service settings in recipient bindings or typed inputs. Current var readings and session history are not package definitions. Builtin selections continue to use the recipient's Raw installation.

Export makes literal var sources and external paths/settings into recipient inputs by default. The CLI has no --include-literals or --include-files flags. For deliberately shared constants or owned files, edit the generated source/manifest or use the SDK's explicit includeLiterals and includeFiles options. Review wildcard rules after export; exact local tool identities are rewritten automatically.

Minimal package with a recipient input

This complete example exports a small agent, a reusable var and a notification hook. Create the three declared paths under a new source directory. The recipient supplies Node on PATH but no model credential is embedded.

raw-package.json:

json
{
  "schema_version": 1,
  "name": "@example/project-kit",
  "version": "1.0.0",
  "description": "An assistant that reads the recipient's project label",
  "files": ["agents/helper.json", "vars/project_label.json", "hooks/notice"],
  "exports": {
    "agents": {"helper": "agents/helper.json"},
    "vars": {"project_label": "vars/project_label.json"},
    "hooks": {"notice": "hooks/notice"}
  },
  "inputs": {
    "type": "object",
    "properties": {"project_label": {"type": "string", "description": "Recipient project label"}},
    "required": ["project_label"]
  },
  "requires": ["raw.agent/1", "raw.hook/2"],
  "metadata": {"external_executables": ["node"]}
}

agents/helper.json (the recipient supplies model):

json
{
  "system_prompt": "Help with the user's project. Read project_label when the project name is relevant. Treat previous variable readings as historical.",
  "tools": {"use": ["builtin/list_vars", "builtin/read_var"]},
  "vars": ["#vars/project_label"],
  "hooks": {"use": ["#hooks/notice"]}
}

vars/project_label.json:

json
{
  "description": "The recipient's project label",
  "access": "read",
  "type": "string",
  "source": {"kind": "literal", "value": {"$input": "project_label"}}
}

The recipient input file for this example is {"project_label":"My project"}. This is a supplied definition value; changing it does not evaluate a provider or template the prompt.

hooks/notice/hook.json and hooks/notice/index.mjs (both covered by files):

json
{"protocol_version": 2, "name":"notice","events":[{"name":"UserPromptSubmit"}],"command":"node","args":["./index.mjs"],"timeout_ms":3000}
js
process.stdin.resume();
process.stdin.on("end", () => process.stdout.write(JSON.stringify({message:"Package hook ran"})));

The hook is selected only by the exported agent. Inspection, installation and binding validate it without running the script; a selected user prompt triggers it. To package a hook alone, export hooks.notice and declare its complete folder, then add pkg/ALIAS/hooks/notice to a direct agent's ordered hooks.use.

Validate, pack and hand off

  1. Run raw package validate DIR and raw package inspect DIR. Resolve missing/undeclared files and invalid references at their source. Validation does not import tools, execute providers, connect MCP or call a model.
  2. Run raw package pack DIR --out FILE.rawpkg, then inspect that archive. Report its name, version, exported components, required inputs and executable prerequisites. There are no install hooks or automatic npm/pip/system dependency installers.
  3. For an artifact request, check the recipient path using an isolated config and data directory as described in references/packages.md. Install the archive, bind the requested export and exercise the relevant harmless operation. A missing external service or model need not block packing; report which live behavior remains unverified. A prose explanation needs no test installation.
  4. Deliver the archive path and exact install/binding commands. Include an input example, a recipient model placeholder and any external prerequisites. The recipient supplies an existing local model alias. Installing alone does not select components or change default_agent.

Send or publish the artifact only when requested. Raw v1 accepts local folders and .rawpkg files; a user can download one from GitHub Releases, but Raw has no hosted marketplace or automatic Git download command.

For a new release, update the source version and repack, then use raw package update ALIAS --from NEW.rawpkg --config CONFIG. Existing sessions can resume with the current runtime. Meaningful prompt/tool changes may rotate Raw's generated cache key once; skill changes can add a tail reload notice. An unchanged next resume is stable; a version label alone is not a reason to reset sessions.

Bundled files

The model reads these on demand while the skill is loaded. They are exposed as readable files and are never executed.

Frequently asked questions

What does the Create Package AI skill do?

Use when packaging or sharing Raw agents, tools, skills, hooks, vars or MCP as a .rawpkg, or preparing an updated release. Covers CLI and dashboard exports, recipient inputs and installation checks.

Why use Create Package on TypingMind?

Because you install it once and use it with any model. Create Package is plain Markdown rather than provider-specific code, so the same skill runs on GPT-5, Claude, Gemini, Grok, or a local model — and you can switch model mid-chat without it breaking. TypingMind runs on your own API keys, so you pay providers directly instead of a per-seat subscription, and your skills and chats stay in your own storage.

How do I install Create Package in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/lploc94/raw-cli/tree/main/examples/skills/create_package. TypingMind reads its SKILL.md and bundles its files and installs it as a skill you can enable per chat.

Which AI models can use Create Package?

Any model you connect in TypingMind. AI skills are plain Markdown instructions rather than provider-specific code, so GPT, Claude, Gemini, Grok, and local models can all load this skill when a request matches it.

How many AI models can I use with Create Package?

As many as you like. As long as a model supports skills, you can use Create Package with it — GPT, Claude, Gemini, Grok, DeepSeek, Mistral, Llama and more — all on TypingMind with your own API keys.

Is the Create Package AI skill free?

Yes. It is published on GitHub by lploc94 under the MIT license. You only pay your own AI provider for the tokens you use.

What are AI skills?

An AI skill is a reusable instruction bundle that teaches an AI model how to do one specific task. It follows the open Agent Skills format: a SKILL.md file with a name and description, plus any scripts, templates or reference files the model may need. The model reads the instructions only when your request matches the skill, so an installed skill costs nothing until it is used.

How are AI skills different from plugins or MCP servers?

A plugin or MCP server gives a model new tools to call — code that runs somewhere and returns a result. An AI skill gives the model knowledge and process instead: how to approach a task, which steps to follow, what good output looks like. Skills are plain Markdown, so they need no server, no API key and no runtime, and they work with any model.

View all

Set up your own AI workspace now

Get notified about new features and future giveaways by subscribing to our newsletter 👇