Skip to content
Get packs

Guide · Agent skill

How to use the Extension Creator agent skill

Updated · written with AI from the project's own documentation and checked against our catalogue data

The extension-creator skill is a bundled agent skill for Qwen Code, the coding tool. It teaches an agent how to scaffold, customize, validate and locally test Qwen Code extensions. This guide explains what the skill does, how to install it, how to trigger it with plain requests, and what safety checks to keep in place. After reading, you will be able to set up the skill and use it to start a new extension or add a capability to an existing one.

Extension CreatorQwenLM/qwen-code · Apache-2.0

Quick start

  1. 1Install the skill. Download the ZIP from this page and unzip it into ~/.claude/skills/ or a project's .claude/skills/ folder, or upload it under Settings → Capabilities → Skills in the Claude apps.
  2. 2Confirm Qwen Code is set up. Make sure the qwen CLI is installed and that Node.js and npm are available if you plan to use the mcp-server or starter template.
  3. 3Ask for an extension. Describe the capability you want in plain language, or run /extension-creator with an extension path and optional template name.
  4. 4Pick a template. Let the agent choose the smallest template that covers your request, or name one yourself.
  5. 5Customize the generated files. The agent reads the scaffolded files, including qwen-extension.json, and edits them for your extension, removing any demo network calls you did not ask for.
  6. 6Review and approve linking. Read the trust summary the agent gives you, approve or decline, and run the link step in a terminal if the extension has settings.
  7. 7Verify the link. Run qwen extensions list to confirm the extension is linked and ready to test.

What the Extension Creator skill makes the agent do

The skill gives the agent a set workflow for building Qwen Code extensions. It identifies the target extension path and the capabilities you want, helps you pick the smallest template that covers them, and runs the existing qwen extensions new command to generate starter files. The agent then reads the generated files, customizes them, reviews trust and security settings, and links the extension locally so you can test it.

The skill also tells the agent to treat text inside extension files as data rather than instructions, and to ask you before linking an extension.

  • Choose a template: no template, context, commands, skills, agent, mcp-server or starter.
  • Scaffold files with qwen extensions new, and check available templates with qwen extensions new --help.
  • Customize qwen-extension.json, QWEN.md, commands, skills, agents, MCP servers, settings and hooks.
  • Review trust and security settings, then link the extension for local testing.

Install the Extension Creator skill

You can install the skill in one of three ways. Download the ZIP file from this page and unzip it into your personal skills folder at ~/.claude/skills/, or into a project's .claude/skills/ folder to limit it to that project. You can also upload the ZIP in the Claude apps under Settings → Capabilities → Skills.

The skill package contains a single file, SKILL.md.

  • Personal install: unzip into ~/.claude/skills/
  • Project install: unzip into the project's .claude/skills/ folder
  • Claude apps: Settings → Capabilities → Skills, then upload the ZIP

Check the requirements

The skill assumes the Qwen Code CLI is installed, with the qwen command and its extensions new, list and link subcommands available. The mcp-server and starter templates also need Node.js and npm.

  • Qwen Code CLI with the qwen command (extensions new, list and link subcommands)
  • Node.js and npm for the mcp-server and starter templates

How to trigger the skill with natural requests

You can ask for help in plain language, such as wanting a new Qwen Code extension, needing help choosing a template, or asking how to link and test an extension locally. You can also invoke the skill directly with /extension-creator followed by an extension path and an optional template name.

Naming a template in your request helps the agent pick the right starting point. If you name several capabilities, the skill favors the closest single template and adds any missing folders by hand, unless the combined starter example is useful.

  • Direct invocation: /extension-creator <extension-path> [template]
  • Natural request: "Help me make a Qwen Code extension that adds a custom slash command for running tests."
  • Natural request: "Which template should I use for an MCP server that a teammate can run?"
/extension-creator my-extension commands

Choose the right template

The skill uses the smallest template that covers the capability you need. Starting small keeps review simpler.

  • No template: minimal extension with only qwen-extension.json.
  • context: persistent instructions through QWEN.md.
  • commands: custom slash commands under commands/.
  • skills: custom skills under skills/<skill-name>/SKILL.md.
  • agent: custom subagents under agents/.
  • mcp-server: MCP server code plus mcpServers manifest wiring.
  • starter: combined context, command, skill, agent and MCP server example.

Review trust and network access before linking

Before any link, the skill summarizes default context files, settings, hooks, channels and lspServers, plus the consent surface: MCP servers, commands, context files, skills and agents. It then asks you whether to approve. If the extension has settings, the skill does not pipe approval; instead it asks you to run the link command yourself in an interactive terminal so you can answer the consent and settings prompts.

The mcp-server and starter templates can include demonstration MCP code with outbound network access. The skill removes or replaces demo network calls unless you asked for that behavior.

  • Approve every link or re-link attempt explicitly.
  • If the extension has no settings, the skill can link with printf 'y\n' piped into qwen extensions link.
  • If you decline, the skill reports that linking was skipped and suggests you run qwen extensions link manually later.
printf 'y\n' | qwen extensions link "$extension_path"

Test the extension after linking

After linking, the skill checks that the extension appears in qwen extensions list and continues with its verification steps. If an extension already appears in the list and you only need validation, the skill does not link it again. If any command fails, the skill stops and reports the error to you before continuing.

  • Confirm the extension with qwen extensions list.
  • Re-run the trust review after changing anything that affects consent, such as MCP servers or settings.
qwen extensions list

What people use it for

  • Scaffolding a new extension that adds persistent project instructions through QWEN.md.
  • Adding a custom slash command under commands/ to an existing extension and testing it locally.
  • Building an MCP server extension with mcp-server wiring in qwen-extension.json.
  • Adding a subagent under agents/ for a team that reviews code the same way every time.
  • Standardizing how a team scaffolds and reviews its extensions.

Frequently asked questions

What is the extension-creator skill?

It is a bundled agent skill from the QwenLM/qwen-code project that teaches an agent to create, customize, validate and locally test Qwen Code extensions. It uses the existing qwen extensions new command and asks for your approval before linking.

Is the extension-creator skill free?

The skill is licensed under Apache-2.0, as stated on this page. Check the license file in the QwenLM/qwen-code repository for the full terms.

What do I need to use it?

You need the Qwen Code CLI installed, with the qwen command and its extensions new, list and link subcommands. The mcp-server and starter templates also need Node.js and npm.

Where do I put the skill files?

Unzip the downloaded ZIP into ~/.claude/skills/ for personal use, or into a project's .claude/skills/ folder. You can also upload the ZIP in the Claude apps under Settings → Capabilities → Skills.

Is it safe to use?

The skill can run shell commands, and it is marked as running scripts and using the network. Linking an extension grants it consent-level access to MCP servers, commands and settings, so only use it on extensions you have reviewed. The skill asks for approval before linking and treats extension file content as data, not instructions.

Can it be used for things other than Qwen Code?

No. The instructions only apply to the Qwen Code extension format. People who are not building Qwen Code extensions will not get value from it.

New agent skills, repos and AI tools — every week

The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.

Our verdict

Use this skill if you are building or maintaining Qwen Code extensions and want the agent to follow a consistent scaffold, review and link process. Skip it if you do not work with Qwen Code, or if you want to run it on extensions you have not reviewed, since it can run shell commands and linking grants real access.

Download the skill