QwenLM/qwen-code

Qwen Code is a programming agent that reads code, modifies features, debugs issues, and runs tests using natural language in the terminal. It also offers IDE plugins, desktop applications, web interfaces, and chat integrations with DingTalk, Feishu, WeChat, and others. The repository homepage recently released Qwen Code 1.0 Preview, indicating that it has open-sourced the combination of Qoder's agent framework capabilities and Qwen Code; during the preview period, interfaces and configurations may still change, and community PRs will only be accepted after 2026 12. The mainline version includes built-in automatic memory, skills, sub-agents, and MCP.

Qwen Code is a programming agent that reads code, modifies features, debugs issues, and runs tests using natural language in the terminal. It also offers IDE plugins, desktop applications, web interfaces, and chat integrations with DingTalk, Feishu, WeChat, and others. The repository homepage recently released Qwen Code 1.0 Preview, indicating that it has open-sourced the combination of Qoder's agent framework capabilities and Qwen Code; during the preview period, interfaces and configurations may still change, and community PRs will only be accepted after 2026 12. The mainline version includes built-in automatic memory, skills, sub-agents, and MCP.

Why it matters A rare combination among open-source programming agents where both the framework and the model are open source. It supports various interfaces including OpenAI, Anthropic, and Gemini, as well as local models, allowing for seamless switching.

Stars 30353 (+64 / 24h)

Reasons

  • Gained 64 stars in the last 24 hours, totaling 30353 stars (indicating attention level only); Apache-2.0 license
  • The new content compared to previous listings is the 1.0 preview version, which has its own branch and documentation

Usage

  • Install using the one-line installation script from the README of the 1.0 preview branch. Run qwen in your project directory, use /auth to log in with a Qwen account, Qoder, or configure third-party interfaces, then use /model to select a model

Risks

  • 1.0 is still in preview; the official documentation states that some interfaces and behaviors may change, so check release notes before upgrading
  • The installation script downloads from Alibaba Cloud Object Storage; model quotas and costs depend on the selected service provider and account

Open

Trendshift(GitHub Trending 快照)

Explore the latest AI Radar updates