qwenlm/qwen-code is an open-source AI coding agent that runs inside the terminal. The name recorded as the most recent release in this repository is sdk-typescript-v0.1.21. The code is written in TypeScript and is published under the Apache-2.0 license. However, the given record contains no changelog for this release, so what was newly added cannot be confirmed.
The core of this repository is a coding CLI that runs locally and directly uses the project's files and tools. The README states that it can be used in the terminal, in editors, on the desktop, in the browser, and in chat, and that it supports Auto-Memory, Auto-Skills, SubAgents, Agent Teams, and MCP. Besides the OpenAI, Anthropic, Gemini, and Qwen APIs and third-party providers, it can also use local models such as Ollama and vLLM. It also provides VS Code, Zed, and JetBrains integrations, a desktop app, a Web UI, an SDK, and Telegram, DingTalk, WeChat, and Feishu channels. GitHub stars are recorded as 28625.
The verifiable evidence extends only to the repository description and the README summary. The given record contains no release notes or change list for sdk-typescript-v0.1.21, and no figures comparing performance. The only filled-in score is a momentum of 5.7 on a 10-point scale, and no freshness, technical, or quality items are given. What the presented range — from terminal to editor, desktop, web, and messenger — means in actual use cannot be confirmed from this material alone. It can be seen as an option that bundles multiple model providers and local models into one CLI, but what it does and how well it does it can only be known by checking the release notes directly.