From qwen
Internal guidance for composing Qwen Code prompts for coding, review, diagnosis, and research tasks inside the Qwen Claude Code plugin
How this skill is triggered — by the user, by Claude, or both
Slash command
/qwen:qwen-promptingThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Use this skill when `qwen:qwen-rescue` needs to ask Qwen for help.
Use this skill when qwen:qwen-rescue needs to ask Qwen for help.
Prompt Qwen like an operator. Compact, block-structured with XML tags. State the task, output contract, default behavior, and extra constraints.
--system-prompt / --append-system-prompt 可塞 schema:比塞在 user prompt 里更稳。mcp_servers 空时不让它摸文件:若 mcp 空数组,prompt 里明确说"只基于我给的 diff,不要读文件"。auto-edit 对非 edit 工具是 auto-deny(v3.1 F-4 实测):无 TTY 环境下 qwen 会生成 permissionDenials,不 hang;prompt 可明确"不要跑 shell 命令" 让它直接产出文本结果。<task>: 具体工作 + 相关仓库/失败上下文<structured_output_contract> or <compact_output_contract>: 精确形状/顺序/简洁要求<default_follow_through_policy>: qwen 默认做什么(而不是问用户)<verification_loop> or <completeness_contract>: debugging/implementation/risky fix 必须<grounding_rules> or <citation_rules>: review / research / 任何可能漂移到无支撑结论的任务completeness_contract + verification_loop + missing_context_gatinggrounding_rules + structured_output_contract + dig_deeper_nudgeresearch_mode + citation_rulesaction_safety(qwen 保持窄,不做无关重构)review / adversarial-review 命令:本身已有 contract。task:diagnosis / planning / research / implementation,需要更细 prompt 控制。task --resume 做跟进:只发 delta 指令,不重复整段 prompt,除非方向实质改变。references/prompt-blocks.md)。<task> 明确 scopeCreates, edits, and optimizes skills for Claude Code, including drafting, evaluating with test prompts, iterating on performance, and improving skill descriptions for better triggering accuracy.
npx claudepluginhub bbingz/qwen-plugin-cc --plugin qwen