qwen-promptinglisted

# Qwen Prompting(qwen3.5-plus 为默认模型;qwen3.6 要 Pro 订阅) Use this skill when `polycli:polycli-provider-agent` 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. ## Core rules - One clear task per Qwen run. Split unrelated asks into separate runs. - Tell Qwen what "done" looks like. Don't assume it will infer. - Prefer explicit output contracts over raising reasoning. - Use XML tags consistently. ## Qwen 3.5/3.6 specifics - **中英混写稳**:可以在 prompt 中中英混用,qwen3.5-plus 对中文 prompt 的处理稳定。 - **`--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。 ## Default prompt recipe - `<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 ## When to add blocks - Coding / debugging: `completeness_contract` + `verification_loop` + `missing_context_gating` - Review / adversarial: `grounding_rules` + `structured_output_contract` + `dig_deeper_nudge` - Research: `research_mo