用 Codex 或 Claude 帮你安装 复制这段 Prompt,粘贴到 Codex、Claude 或其他助手里,让它检查 Skill 页面并帮你完成安装。
直接命令不会经过审查 Prompt;运行前请先检查来源。
npx skills add https://github.com/Dev-Toolbelt/dev-team-agents --skill interaction-patterns命令会保持在同一行。复制前请横向滚动并检查完整内容。
想先保存到本地?可下载 SkillsMP 当前能够提供的文件。
正在显示 SKILL.md
基于 SOC 职业分类
| name | interaction-patterns |
| description | Quiz-first rule — use AskUserQuestion for any user-facing choice, never plain text. |
Whenever you need to ask the user a question that has a finite set of reasonable answers, use the AskUserQuestion tool instead of plain text.
This applies to:
| Situation | Use AskUserQuestion? |
|---|---|
| Yes / No confirmation | ✅ Always |
| Multiple-choice selection (2–4 options) | ✅ Always |
| Picking between named alternatives (branch vs worktree, module by module vs all at once) | ✅ Always |
| Asking for a free-form name, title, or description with no obvious set of values | ❌ Ask as plain text |
| Reporting status or results (no question involved) | ❌ Plain output |
Include an "Other" option (with description: "Type your own answer.") whenever:
Do not add "Other" when:
Use the same language as the language field in .dev-team-agents/user-data/preferences.json for:
question textheader labellabel and description of each optionFall back to English if the file is unreadable.
{
"questions": [
{
"question": "<question text>?",
"header": "<short label, max 12 chars>",
"multiSelect": false,
"options": [
{ "label": "Yes", "description": "<what happens if yes>" },
{ "label": "No", "description": "<what happens if no>" }
]
}
]
}
{
"questions": [
{
"question": "<question text>?",
"header": "<short label>",
"multiSelect": false,
"options": [
{ "label": "<Option A>", "description": "<explain A>" },
{ "label": "<Option B>", "description": "<explain B>" },
{ "label": "<Option C>", "description": "<explain C>" },
{ "label": "Other", "description"
{
"questions": [
{
"question": "<question text>?",
"header": "<short label>",
"multiSelect": true,
"options": [
{ "label": "<Option A>", "description": "<explain A>" },
{ "label": "<Option B>", "description": "<explain B>" },
{ "label": "<Option C>", "description": "<explain C>" }
]
}
]
}
{
"questions": [
{
"question": "Should I work in an isolated git worktree for this task?",
"header": "Isolation",
"multiSelect": false,
"options": [
{ "label": "Yes — worktree", "description": "Create an isolated worktree. I'll ask for the base branch next." },
{ "label": "No — new branch", "description": "Stay in the current worktree and check out a new branch." }
]
}
]
}
{
"questions": [
{
"question": "This task covers a large scope and may consume significant tokens. How do you want to proceed?",
"header": "Scope",
"multiSelect": false,
"options": [
{ "label": "Module by module", "description": "Process one module at a time with a checkpoint between each (safer)." },
{ "label": "All at once", "description": "Process the entire scope in a single pass (faster, higher token cost)." },
{ "label": "Cancel", "description": "Stop here — do not proceed." }
]
}
]
{
"questions": [
{
"question": "Pre-commit checks failed. How do you want to proceed?",
"header": "Gate failure",
"multiSelect": false,
"options": [
{ "label": "Fix and re-stage", "description": "Fix the issues now, re-stage, and retry the commit." },
{ "label": "Commit anyway", "description": "Skip the failing checks and commit as-is." },
{ "label": "Abort", "description": "Cancel the commit entirely." }
]
}
]
}
Presented after plan approval, before execution begins:
{
"questions": [
{
"question": "How should this plan be executed?",
"header": "Exec Strategy",
"multiSelect": false,
"options": [
{ "label": "Isolated worktree (Recommended)", "description": "Execute in a git worktree isolated from the main development tree, with Docker isolation if enabled in preferences." },
{ "label": "New branch", "description": "Create a new branch from the current one and execute there." },
{ "label": "Current branch", "description": "Execute directly on the current branch without any git changes." },
{ "label"
The first option MUST be the recommended one based on preferences. See plan-mode/SKILL.md → Execution Strategy Gate for the full protocol.
{
"questions": [
{
"question": "Plan ready. Do you approve?",
"header": "Plan approval",
"multiSelect": false,
"options": [
{ "label": "Approved — proceed", "description": "Execute the plan as described." },
{ "label": "Adjust first", "description": "I'll describe what to change before you proceed." },
{ "label": "Cancel", "description": "Discard the plan and stop." }
]
}
]
}