Category: provider
Model Studio Qwen Coder
Validation
CODEBLOCK0
Pass criteria: command exits 0 and output/aliyun-qwen-coder/validate.txt is generated.
Output And Evidence
- Save prompts, repository context summaries, and normalized coding request payloads under output/aliyun-qwen-coder/. Record the exact model, endpoint mode, and target language/framework for reproducibility.
Use Qwen Coder for coding assistants, code review drafting, repository-level reasoning, patch planning, and tool-using software agents.
Critical model names
Use one of these exact model strings as appropriate:
- INLINECODE2 INLINECODE3 INLINECODE4 INLINECODE5
Prerequisites
- Install SDK in a virtual environment:
CODEBLOCK1
- Set DASHSCOPE_API_KEY in your environment, or add dashscope_api_key to ~/.alibabacloud/credentials. Prefer the OpenAI-compatible endpoint when a client does not yet expose the latest coder models in dropdown UI.
Normalized interface (code.generate)
Request
- messages (array, required) INLINECODE10 (string, optional): default INLINECODE11 INLINECODE12 (string, optional) INLINECODE13 (array, optional) INLINECODE14 (string, optional) INLINECODE15 (array, optional) INLINECODE16 (bool, optional)
Response
- text (string) INLINECODE18 (string, optional) INLINECODE19 (object, optional)
Quick start
CODEBLOCK2
Operational guidance
- Pass only the files relevant to the requested change to reduce noise. Use qwen3-coder-next for current-generation coding tasks and qwen3-coder-plus when you need a stronger but potentially costlier coder. For repo-scale changes, include architecture notes and expected test commands. Prefer deterministic prompts and pinned model IDs for benchmarking or regression comparison.
Output location
- Default output: INLINECODE22 Override base dir with OUTPUT_DIR.
References
技能名称: aliyun-qwen-coder
详细描述:
类别: 提供方
Model Studio Qwen Coder
验证
bash
mkdir -p output/aliyun-qwen-coder
python -m pycompile skills/ai/code/aliyun-qwen-coder/scripts/prepare coderequest.py && echo py compile_ok > output/aliyun-qwen-coder/validate.txt
通过标准: 命令退出码为0,且生成了 output/aliyun-qwen-coder/validate.txt 文件。
输出与证据
- 将提示词、仓库上下文摘要以及标准化编码请求负载保存到 output/aliyun-qwen-coder/ 目录下。 记录确切的模型、端点模式以及目标语言/框架,以确保可复现性。
使用 Qwen Coder 进行编码助手、代码审查草稿、仓库级推理、补丁规划以及工具型软件代理。
关键模型名称
根据情况使用以下确切的模型字符串之一:
- qwen3-coder-next qwen3-coder-plus qwen-coder-plus qwen2.5-coder-32b-instruct
前提条件
bash
python3 -m venv .venv
. .venv/bin/activate
python -m pip install dashscope
- 在你的环境中设置 DASHSCOPEAPI KEY,或者将 dashscopeapi key 添加到 ~/.alibabacloud/credentials 文件中。 当客户端在下拉界面中尚未暴露最新的编码器模型时,优先使用兼容 OpenAI 的端点。
标准化接口 (code.generate)
请求
- messages (数组<对象>,必填) model (字符串,可选):默认为 qwen3-coder-next repository_summary (字符串,可选) files (数组<字符串>,可选) language (字符串,可选) tools (数组<对象>,可选) stream (布尔值,可选)
响应
- text (字符串) patch (字符串,可选) usage (对象,可选)
快速开始
bash
python skills/ai/code/aliyun-qwen-coder/scripts/preparecode request.py \
--task 将请求验证重构为一个小助手,并添加一个单元测试。 \
--language python
操作指南
- 仅传递与请求变更相关的文件,以减少噪音。 对于当前世代的编码任务,使用 qwen3-coder-next;当需要更强但可能成本更高的编码器时,使用 qwen3-coder-plus。 对于仓库级别的变更,请包含架构说明和预期的测试命令。 在进行基准测试或回归比较时,优先使用确定性的提示词和固定的模型 ID。
输出位置
- 默认输出:output/aliyun-qwen-coder/requests/ 通过 OUTPUT_DIR 覆盖基础目录。
参考资料