@lynn123411/dsh-ask-user-grilling 0.1.1 → 0.2.0

package/README.md

@@ -1,15 +1,16 @@
 # @lynn123411/dsh-ask-user-grilling
 
-DSH 侧的 grilling 适配层（输送机制）：把 Matt Pocock 的 grilling 流程在 DSH 里的提问环节做成工具级硬约束。**不修改、不包装这组技能的任何文件**——技能按原设计运行，本插件只负责「在 DSH 里怎么问」。
+DSH 侧的 grilling 适配层（输送机制）：把 Matt Pocock 的 grilling 流程在 DSH 里的提问环节做成工具级硬约束。本插件只提供 `ask_user_grilling`，不提供任何 plan-mode 工具——共识达成后不自动进入 plan mode，交还用户决定下一步。分工：本插件负责「在 DSH 里怎么问」的工具与报错层；grilling 纪律文案（强制走 `ask_user_grilling`、子代理停轮）写在工具描述与 matt-* 预设 vendor 的 `skills/grilling/SKILL.md` 本地适配旁注（DSH delivery、Sub-agent rounds，重同步上游技能时需保留）里，preset persona 保持原厂逐字不做任何修改。
 
 ## 特性
 
 - **ask_user_grilling**：grilling 轮次专用提问工具。
   - **子代理闸门**：后台有子代理运行时会拒绝提问，返回「阻塞 + 运行中名单」——agent 应**结束当前回合**，子代理结算通知自动唤醒后再调用，不在回合内反复重试；
-  - **强制多选**：所有问题一律多选（schema 不提供关闭开关）；
-  - **补充机制**：自动为每个问题追加「补充」选项，并自动在每轮末尾追加「本轮还有什么要补充或调整的吗？」；
-  - **题干引导**：要求题干只含问题本身、不重复选项文本（仅模型侧引导，不做硬校验——避免误伤自然提及选项名称的题干）。
-- **enter_plan_mode**：为当前 agent 激活 DSH 计划模式。grilling 最后一轮答完、用户确认共识后调用，让 agent 写方案供审阅而不是直接开始执行；可通过 `exit_plan_mode` 或 `/plan off` 退出。
+  - **强制多选**：所有问题一律多选（schema 不提供关闭开关；此行为刻意**不写入工具描述**——模型若知道只能多选，会为避免互斥选项而影响出题质量，见「描述即纪律」）；
+  - **补充机制**：每题末尾的补充输入框由 UI 自动渲染、轮末补充题由代码自动追加，两者都不依赖模型也不写入描述（模型自加补充项只会与它们重复）；仅当轮末补充输入非空时才应再开一轮；
+  - **题干引导**：要求题干只含问题本身、不重复选项文本（仅模型侧引导，不做硬校验——避免误伤自然提及选项名称的题干）；
+  - **描述即纪律**：工具描述保持精简，只承载「grilling 轮次专用（其余用 ask_user_question）、先散文预告同一轮、再以一次调用投递表单、字段映射、勿自加收尾题、子代理运行中返回 blocked」等工具必知项；投递协议细节（散文预告与表单一一对应、PTC 形态）由技能旁注（DSH delivery / Sub-agent rounds）承载，不与工具描述重复。
+- 轮次收尾不自动进入 plan mode：grilling 达成共识后由用户决定继续方式（直接执行、或需要方案时自行 `/plan on`）。
 
 ## 安装
 

package/lib/index.js

@@ -9,28 +9,22 @@ import "@deepseek-ai/dsh-user-questions";
  * ask_user_grilling:
  *   - gate: refuses while background subagents are running (R1)
  *   - forces multi-select on every question (R2)
- *   - appends a per-question "Supplement" option and a round-end question (R3)
+ *   - appends a round-end supplement question (R3) — per-question supplement is
+ *     via the built-in custom input ("Type your answer" / "输入你的答案"), no
+ *     extra per-question "Supplement" option to avoid duplication with that
+ *     field (see image.png issue: checkbox + input were redundant)
  *   - stem/option separation is guidance only — never rejects stems (R4
  *     relaxed: substring checks false-positive on legitimate stems)
- *
- * enter_plan_mode:
- *   - activates DSH plan mode for the current agent (planMode.set)
  */
 const name = "tool-ask-user-grilling";
 const inject = ["tools", "userQuestions"];
 
-const SUPPLEMENT_OPTION = {
-  label: "Supplement",
-  description: "Select this and write your supplement for this question in your reply.",
-};
-
 const ROUND_END_QUESTION = {
   id: "__grill_round_supplement__",
-  question: "Anything else to add or adjust this round?",
-  header: "Round supplement",
+  question: "这轮还有什么要补充或调整的吗？",
+  header: "轮末补充",
   options: [
-    { label: "Nothing to add" },
-    { label: "I have something to add (write it in my reply)", description: "Select this and write any additions or adjustments for this round in your reply." },
+    { label: "无需补充" },
   ],
   multiSelect: true,
 };
@@ -42,12 +36,12 @@ function displayName(entry) {
 function apply(ctx) {
   ctx.tools.register(defineTool({
     name: "ask_user_grilling",
-    description: "Ask the user a ROUND of grilling questions (decision-tree interview). Use for grilling sessions (grill-me / grill-with-docs / triage / wayfinder / architecture grilling): send ALL frontier questions of the current round in ONE call. The tool forces multi-select, appends a per-question 'Supplement' option, and appends a round-end supplement question automatically. If background subagents are running, this tool returns blocked — do NOT retry within the same turn: end your turn and wait; the settlement notice will wake you automatically, then call again. Keep the question stem to the question itself, without repeating option labels (style guidance only — stems are never rejected). Example of one round with two questions: questions: [{ id: 'q1', question: 'Which issue tracker?', options: [{ label: 'GitHub' }, { label: 'Local markdown' }] }, { id: 'q2', question: 'Any deadline?', options: [{ label: 'This week' }, { label: 'Next month' }] }].",
+    description: "Deliver one ROUND of grilling questions as a form. Use it only when the grilling skill (grill-me / grill-with-docs, or the grilling phases of triage / wayfinder / improve-codebase-architecture) directs a round: first announce the whole round in the message text (title, options and your recommendation, per the skill's template), then deliver the SAME round as ONE call here — the prose and the form must match one-to-one. Map each question to the fields below (title → header, body → question, the A/B/C choices → options; put your recommendation first in options with \"(Recommended)\"; if your recommendation is not an option, state it briefly in the question text). Each question needs a stable id that does not start with __grill_ (reserved for the auto-appended round-end supplement question — never add your own catch-all/\"anything else?\" question; a non-empty supplement input reshapes the tree: ask a further round, and stop asking once the user confirms shared understanding). If background subagents are still running, this tool returns blocked: end your turn and wait for the settlement notice, do not retry within the same turn. For any non-grilling question use the plain ask_user_question tool.",
     parameters: {
       questions: {
         type: "array",
         required: true,
-        description: "All frontier questions of the current round, each with a stable id, the question text (stem) and options.",
+        description: "The current round's questions (round/frontier protocol: see the grilling skill). Each needs a stable id, a stem and options.",
         items: {
           type: "object",
           additionalProperties: true,
@@ -60,15 +54,15 @@ function apply(ctx) {
             question: {
               type: "string",
               required: true,
-              description: "The question stem only — never include option text here.",
+              description: "The question stem — write only the question. The A/B/C choices belong in options, never in the stem.",
             },
             header: {
               type: "string",
-              description: "Optional short heading for the question, such as \"Round 3\".",
+              description: "Optional short heading, e.g. \"Q2 — Deadline\".",
             },
             options: {
               type: "array",
-              description: "Choices to show the user. If you recommend one, put it first and append \"(Recommended)\" to that label.",
+              description: "Choices to show the user. Put your recommended option first and append \"(Recommended)\" to its label.",
               items: {
                 type: "object",
                 additionalProperties: true,
@@ -112,16 +106,32 @@ function apply(ctx) {
             items: { type: "string" },
             description: "Validation violations when rejected (reserved id prefix only).",
           },
-          error: { type: "string" },
+          error: {
+            type: "string",
+            description: "Human-readable error when blocked or rejected.",
+          },
           answers: {
             type: "array",
+            description: "One entry per question, in the order asked.",
             items: {
               type: "object",
               additionalProperties: false,
               properties: {
-                id: { type: "string", required: true },
-                selected: { type: "array", required: true, items: { type: "string" } },
-                custom: { type: "string" },
+                id: {
+                  type: "string",
+                  required: true,
+                  description: "The question id you supplied.",
+                },
+                selected: {
+                  type: "array",
+                  required: true,
+                  items: { type: "string" },
+                  description: "Labels of the options the user picked (may be empty if skipped).",
+                },
+                custom: {
+                  type: "string",
+                  description: "User-typed free text for this question, if any.",
+                },
               },
             },
           },
@@ -149,7 +159,7 @@ function apply(ctx) {
           return {
             blocked: true,
             waiting: [],
-            error: "Cannot confirm background subagent status (subagents query failed). Please call this tool again later.",
+            error: "Cannot confirm background subagent status (subagents query failed). End your turn and wait; call this tool again in a later turn.",
           };
         }
       }
@@ -180,7 +190,7 @@ function apply(ctx) {
         };
       }
 
-      // 3. transform: force multi-select (R2), append supplement option (R3)
+      // 3. transform: force multi-select (R2); per-question supplement is via the built-in custom input ("Type your answer"/"输入你的答案") — no extra option is added to avoid duplication with that field
       const questions = args.questions.map((question) => ({
         id: question.id,
         question: question.question,
@@ -190,12 +200,11 @@ function apply(ctx) {
             label: option.label,
             ...(option.description !== undefined ? { description: option.description } : {}),
           })),
-          SUPPLEMENT_OPTION,
         ],
         multiSelect: true,
       }));
 
-      // 4. round-end supplement question (R3)
+      // 4. round-end supplement question (R3) — single "无需补充" option; supplement is via custom input, so no "I have something to add" option (duplicates that field)
       questions.push(ROUND_END_QUESTION);
 
       // 5. ask through the userQuestions seam (UI renders from the service)
@@ -213,35 +222,6 @@ function apply(ctx) {
       };
     },
   }));
-
-  ctx.tools.register(defineTool({
-    name: "enter_plan_mode",
-    description: "Enter plan mode for the current agent. Call it after the final round of a grilling session once the user has confirmed shared understanding, so a plan is written for review instead of executing directly. Plan mode ends via exit_plan_mode or the /plan off command.",
-    parameters: {},
-    output: {
-      schema: {
-        type: "object",
-        additionalProperties: false,
-        properties: {
-          ok: { type: "boolean" },
-          result: { type: "string" },
-          error: { type: "string" },
-        },
-      },
-      render: (_args, value) => [{ type: "text", text: JSON.stringify(value) }],
-    },
-    async execute(_args, exec) {
-      const planMode = ctx.get("planMode");
-      if (planMode === undefined) return { ok: false, error: "planMode service unavailable" };
-      const agent = exec.agent;
-      if (agent === undefined) return { ok: false, error: "current agent unavailable" };
-      const outcome = planMode.set(agent, true);
-      // cancelled = an opposite pending selection was cleared and the logged
-      // state already matches the requested active state — plan mode IS active.
-      const ok = outcome === "committed" || outcome === "queued" || outcome === "cancelled" || outcome === "noop";
-      return { ok, result: `plan mode ${outcome}` };
-    },
-  }));
 }
 
 export { apply, inject, name };

package/package.json

@@ -1,7 +1,7 @@
 {
   "name": "@lynn123411/dsh-ask-user-grilling",
-  "version": "0.1.1",
-  "description": "DSH delivery adaptations for grilling rounds: ask_user_grilling (subagent gate / forced multi-select / supplement option / stem guidance) and enter_plan_mode. Does not modify any skill file.",
+  "version": "0.2.0",
+  "description": "DSH delivery adaptations for grilling rounds: ask_user_grilling (subagent gate / forced multi-select / round-end supplement / stem guidance). Does not modify any skill file.",
   "type": "module",
   "main": "lib/index.js",
   "exports": {