Qwen MCP Tool

You can run and connect to the Qwen MCP Tool as an MCP server to empower AI assistants with Qwen’s large-context analysis, file-aware prompts, sandboxed execution, and model flexibility. This server exposes a straightforward way to invoke Qwen via MCP clients, manage files within prompts, and safely run tests in isolated environments.

How to use

Connect your MCP client to the Qwen MCP Server using the provided stdio configuration. Once connected, you can ask Qwen to analyze code, summarize architectures, or run tests with optional sandboxing and model selection. Use file references like @filename or @directory to include file contents in your queries, and choose an approval mode to control how tool calls are executed.

How to install

claude mcp add qwen-cli -- npx -y @jaggerxtrm/qwen-mcp-tool

This single command configures everything automatically within Claude Code for MCP usage.

If you prefer to install the MCP tool globally and configure it manually, follow these steps to install and prepare the MCP server for use with Claude Code.

npm install -g @jaggerxtrm/qwen-mcp-tool

Then add to Claude Code MCP settings to expose the server under the name `qwen-cli` with the default command configuration starting the MCP tool. The configuration format is a JSON object where the `mcpServers` entry maps a short name to its command. For example, you can place the following in your MCP settings:

{ "mcpServers": { "qwen-cli": { "command": "qwen-mcp-tool" } } }

## From source development

git clone <repo-url> cd qwen-mcp-tool npm install

Build the project and set it up for local development.

Notes on setup flow

If you want to use the MCP server via a direct CLI installer within Claude, the recommended install path is the quick setup command shown above. Otherwise, you can install globally and configure the MCP settings JSON manually or use npx for an inline configuration.

What you can do with the tool

• Analyze code with large context windows for files and directories using @filename or @directory prompts.

• Run code safely in a sandbox to test hypotheses or verify behavior without risking your environment.

• Switch between Qwen models (for example, qwen3-coder-plus and qwen3-coder-turbo) depending on your needs and quotas.

• Control how tool calls are handled with multiple approval modes such as plan, default, auto-edit, and yolo to tailor the interaction flow.

Configuration and usage tips

Default models are:

• Primary: qwen3-coder-plus
• Fallback: qwen3-coder-turbo (used if the primary hits quota limits) You can override these by specifying the model parameter in your requests.

Troubleshooting

If you encounter issues such as the Qwen CLI not being found or timeouts when handling large files, consider installing the Qwen CLI and adjusting your query strategy. Use smaller chunks or enable plan mode to analyze without executing.

Security and practices

Use sandbox mode for any code execution to minimize risk, and apply appropriate approval modes to ensure you retain control over what the tool can do within your environment.

Examples

{"prompt": "@src/main.ts Explain what this code does"}

{"prompt": "@src/ Summarize the architecture of this codebase"}

{"prompt": "Run the test suite and fix any failures", "model": "qwen3-coder-plus", "sandbox": true, "approvalMode": "auto-edit"}

Available tools

ask-qwen

Main MCP tool for interacting with Qwen AI. Supports file references (@filename, @directory), model selection, sandbox mode, and multiple approval modes.

ping

Simple echo tool to verify that the MCP connection is alive.

Help

Display help information for available commands and usage options.