@ai-setting/roy-task-plugin 0.7.10 → 0.7.12

package/dist/cli.js

@@ -456,6 +456,62 @@ import {
   TASK_DISCIPLINE_LIFECYCLE,
   TASK_DISCIPLINE_OPERATION
 } from "@ai-setting/roy-agent-core/env/task/prompt";
+
+// ../agent/src/trim-permission.ts
+var TRIM_RULES = Object.freeze([
+  {
+    name: "complete-禁止调用",
+    match: (line) => /不要调用/.test(line) && /task_complete/.test(line) && /权限/.test(line)
+  },
+  {
+    name: "track-禁止置终态",
+    match: (line) => /终态由验收方落/.test(line) || /不要/.test(line) && /置为/.test(line) && /completed/.test(line)
+  },
+  {
+    name: "lifecycle-角色二分标题",
+    match: (line) => /收尾分两种角色/.test(line),
+    replace: () => "6. COMPLETE — 收尾：用 `task_complete` after item-by-item verification",
+    dropFollowing: 5
+  },
+  {
+    name: "complete-标题角色限定",
+    match: (line) => /^#{1,6}\s/.test(line) && /仅验收方适用/.test(line),
+    replace: (line) => line.replace(/\s*（\*\*仅验收方适用\*\*）/, "")
+  },
+  {
+    name: "complete-角色作用域引用块",
+    match: (line) => /角色作用域/.test(line) || /若你是/.test(line) && /执行方/.test(line) && /权限/.test(line) || /你的收尾动作是/.test(line) && /验收由主智能体负责/.test(line)
+  }
+]);
+function trimTaskCompletePermissionText(text) {
+  if (!text)
+    return text;
+  const lines = text.split(`
+`);
+  const out = [];
+  for (let i = 0;i < lines.length; i++) {
+    const line = lines[i];
+    const rule = TRIM_RULES.find((r) => r.match(line));
+    if (!rule) {
+      out.push(line);
+      continue;
+    }
+    if (rule.replace)
+      out.push(rule.replace(line));
+    const drop = rule.dropFollowing ?? 0;
+    for (let k = 0;k < drop && i + 1 < lines.length; k++) {
+      i++;
+    }
+  }
+  return out.join(`
+`);
+}
+
+// ../agent/src/roy-prompt-core.ts
+var TRIMMED_LIFECYCLE = trimTaskCompletePermissionText(TASK_DISCIPLINE_LIFECYCLE);
+var TRIMMED_GOALS = trimTaskCompletePermissionText(TASK_DISCIPLINE_GOALS);
+var TRIMMED_OPERATION = trimTaskCompletePermissionText(TASK_DISCIPLINE_OPERATION);
+var TRIMMED_COMPLETE = trimTaskCompletePermissionText(TASK_DISCIPLINE_COMPLETE);
 var ROY_PROMPT_COMMON = `# Roy — task-first-class-citizen agent
 
 You are **Roy**, an agent whose defining rule is simple:
@@ -465,16 +521,16 @@ You are **Roy**, an agent whose defining rule is simple:
 You are a worker agent: you act through the tools, and your output is what
 the user sees.
 
-${TASK_DISCIPLINE_LIFECYCLE}
+${TRIMMED_LIFECYCLE}
 
-${TASK_DISCIPLINE_GOALS}
+${TRIMMED_GOALS}
 
 The tool returns a **numeric task id** — that number is your handle for the
 rest of the conversation.
 
-${TASK_DISCIPLINE_OPERATION}
+${TRIMMED_OPERATION}
 
-${TASK_DISCIPLINE_COMPLETE}
+${TRIMMED_COMPLETE}
 
 #### Several deliverables in one field
 \`deliverable_storage_path\` may name **multiple paths**, space-separated
@@ -510,7 +566,78 @@ if it is not in the task store, it is not tracked.
 - Report the task id and its state whenever you finish a step.
 - If a task is blocked, mark it (\`task_update(status="paused")\`) and record a
   \`problem\` milestone explaining the blocker instead of going silent.`;
+var ROY_PROMPT_CLAUDE_DISCIPLINE = `
+## Milestone discipline (Claude runtime)
+
+The task card renders **your** milestones. A run that calls tools but
+records nothing leaves the card empty, and the user sees a failed task with
+no trace of the work. Measured failure: 67 tool calls, 2 automatic hooks,
+0 real milestones, then the run was killed and the card froze in a failed
+state. Read this section as binding.
+
+### Record per stage, immediately
+
+- Every time you finish a **阶段** — a sub-goal, an investigation, a file
+  written, a test that passed, a decision, a blocker — call
+  \`task_operation_create\` **立即**, in that same turn.
+- Do **not** batch. A milestone written after six tool calls reads as
+  retrospective filler; the point is that the card shows the work *as it
+  happens*.
+- If a single tool call produced a stage-level result, that call is
+  followed by exactly one milestone — no more, no less.
+- Long-running work (> 5 minutes) still records as it goes, exactly as the
+  shared contract requires.
+
+### The shape of a milestone
+
+\`milestone_title\` — a **短语**, ≤ 20 字, verb-first, naming what was
+locked down. Not a sentence, not a label like "progress".
+
+- ✅ \`泄漏源锁定\`, \`测试全绿\`, \`缓存层落地\`, \`回滚路径确认\`
+- ❌ \`进度更新\`, \`继续处理\`, \`做了一些工作\`, \`阶段完成\`
+
+Those ❌ examples are **空壳** milestones: they carry no information about
+what changed, so they are worse than useless — they make the card look
+busy while hiding the work. Never write a title like those.
+
+\`milestone_description\` — a dense **技术叙述** of the state you reached.
+Name the concrete things, so a reader can reconstruct the work without
+re-running it:
+
+- the **文件路径** you wrote or read
+- the **函数名** / symbol you changed or traced
+- the **命令** you ran and its result
+- the numbers: pass/fail counts, line counts, exit codes, sizes, timings
+
+Example (title + description):
+
+\`\`\`
+milestone_title: "泄漏源锁定"
+milestone_description: "test_workflow_error_handling 的生成器被放弃 → registry 由 GC 延迟回收；泄漏路径锁定在 WorkflowRegistry.dispose()，packages/core/src/registry.ts:412"
+\`\`\`
+
+\`process_description\` (100–500 字符) — how it was reached (**如何达到**):
+the strategy and the tool steps. It must **not** repeat the description —
+the description is the *state*, the process is the *route*. Repeating
+one in the other is a second ❌.
+
+### Tool-call budget
+
+If you have made several **连续的工具调用** without recording a
+milestone, the next action is \`task_operation_create\` — not another
+tool call. Two, three tool calls in a row is normal exploration; a long
+streak without a milestone is the failure mode measured above. When the
+streak is getting long, write down what you have already established, then
+continue.
+
+### What good looks like
+
+For a task with 3–4 stages you should end up with **3–4 hand-written
+milestones** in the card — not 2 automatic hooks and nothing else. The
+automatic \`收到消息\` / \`Task 完成\` hooks are not a substitute for yours.
+`;
 var ROY_PROMPT_CLAUDE_TAIL = `
+${ROY_PROMPT_CLAUDE_DISCIPLINE}
 ## Tool surface on Claude Code
 
 The \`task_*\` tools are served by the **\`roy-task\` MCP server** this runtime
@@ -1084,6 +1211,7 @@ function errMsg(err) {
 
 // ../../../packages/cli/src/daemon/discipline.ts
 import { TASK_DISCIPLINE_PROMPT } from "@ai-setting/roy-agent-core/env/task/prompt";
+var PLUGIN_TASK_DISCIPLINE_PROMPT = trimTaskCompletePermissionText(TASK_DISCIPLINE_PROMPT);
 var RUNTIMES_WITH_AGENT_SYSTEM_PROMPT = Object.freeze([
   "opencode",
   "claude"
@@ -1152,7 +1280,7 @@ function buildCodexArgs(input) {
   return args;
 }
 function withDiscipline(prompt, systemPrompt) {
-  const contract = systemPrompt === undefined ? TASK_DISCIPLINE_PROMPT : systemPrompt;
+  const contract = systemPrompt === undefined ? PLUGIN_TASK_DISCIPLINE_PROMPT : systemPrompt;
   if (!contract)
     return prompt;
   if (prompt.includes(contract))
@@ -1692,7 +1820,7 @@ function buildDispatchPrompt(input) {
     ""
   ];
   if (needsDispatchInjection(input.runtime)) {
-    parts.push("## 任务纪律（Task #4889 — 来自 roy-agent core 单一真相源）", "", TASK_DISCIPLINE_PROMPT, "");
+    parts.push("## 任务纪律（Task #4889 — 来自 roy-agent core 单一真相源）", "", PLUGIN_TASK_DISCIPLINE_PROMPT, "");
   }
   parts.push("## 硬性规则", "", "1. **不要**自己调用 `lark-cli` 发送回复。守护进程会在你结束后，把你的", "   **最终回答文本**用 `lark-cli im +messages-reply` 一次性发回飞书。", "   你自己发会造成重复消息。", "2. 把**最终回答**写成给发件人看的**普通散文 / Markdown**。", "   不要输出 JSON 包裹、不要输出“我调用了某某工具”的过程流水账。", "   这段文本会**原样**成为飞书里的回复正文。", "3. 需要落任务就用 `task_*` 工具（task_create / task_update / …），", "   这是你的主要职责；不要只在回答里“描述”你打算做什么。", "", "## 什么时候该建任务（Task #4433）", "", "任务由**你**决定建不建，守护进程不会替你建。判断标准：", "", "- **该建**：需要多步操作、要动文件/跑命令、要调研并产出结论、", "  耗时预计超过一两分钟、或用户明确要求「跟踪 / 建个任务 / 后续跟进」。", "  → 用 `task_create` 建任务，**然后**再开始干活（先立任务，再动手）。", "- **不必建**：寒暄、谢谢、纯问答（「今天几号」「这个参数叫啥」）、", "  一两句话就能答完且不需要任何工具的问题。", "  → 直接回答，不要为了「显得勤快」而建一个空任务。", "- **建完要更新**：动手过程中用 `task_update` 推进度 / 改 current_status，", "  干完用 `task_complete` 收尾。建了任务却从不更新，比不建更糟——", "  看板上会留下一个卡住的任务。", "", "⚠️ **卡片与任务一一对应**：只有你建了任务，用户才会收到一张进展卡片；", "没建任务就只有文字回复。所以**不要**为了「想要一张卡片」而去建任务，", "也不要建了任务又不用它——卡片上显示的就是你建的那个任务的真 id 和标题。", "", "## 环境事实（必须使用下面这些值，不要自己猜）", "", `- lark-cli profile：\`${profile}\`（飞书 openId 是 per-app 隔离的，换 profile 会失败）`, `- chat_id：\`${chatId}\``, `- message_id：\`${messageId}\``, `- event_id：\`${eventId}\``, "", "## 原始事件（完整 JSON）", "", "```json", JSON.stringify(input.rawEvent, null, 2), "```");
   if (input.recommendedAction) {
@@ -1710,6 +1838,25 @@ function buildOpencodeArgs(input) {
 }
 
 // ../../../packages/cli/src/daemon/task-card.ts
+async function hydrateCardTimeline(notifier, input) {
+  try {
+    if (typeof notifier.hydrateState !== "function")
+      return;
+    const taskId = taskIdOf(input.meta);
+    if (taskId === undefined)
+      return;
+    const store = await loadLedger();
+    if (!store)
+      return;
+    const record = store.get(taskId);
+    if (!record?.messageId || !record.timeline)
+      return;
+    if (record.messageId !== input.messageId)
+      return;
+    const snapshot = JSON.parse(record.timeline);
+    notifier.hydrateState(record.messageId, snapshot);
+  } catch {}
+}
 var CARD_OP_TIMEOUT_MS = 15000;
 
 class CardTimeoutError extends Error {
@@ -1735,11 +1882,14 @@ function withTimeout(op, promise, ms) {
   });
 }
 var ledgerStores = new Map;
+var ledgerOverride;
 var ledgerInFlight;
 function ledgerPathKey() {
   return process.env.ROY_AGENT_LARK_CARD_DB ?? "";
 }
 async function loadLedger() {
+  if (ledgerOverride)
+    return ledgerOverride();
   const key = ledgerPathKey();
   if (ledgerStores.has(key))
     return ledgerStores.get(key);
@@ -1826,6 +1976,8 @@ function createRoyAgentCardPort(options = {}) {
   const opTimeoutMs = options.opTimeoutMs ?? CARD_OP_TIMEOUT_MS;
   let notifier;
   let inFlight;
+  if (options.loadLedger)
+    ledgerOverride = options.loadLedger;
   const loadNotifier = async () => {
     const mod = await import("@ai-setting/roy-agent-core/env/task/plugins");
     if (typeof mod.LarkCardPatchNotifier !== "function") {
@@ -1879,6 +2031,7 @@ function createRoyAgentCardPort(options = {}) {
     },
     async finish(input) {
       const n = await withTimeout("load", getNotifier(), opTimeoutMs);
+      await hydrateCardTimeline(n, input);
       await withTimeout("finish", n.finishTask({
         messageId: input.messageId,
         chatId: input.chatId,
@@ -3406,7 +3559,7 @@ function formatStatus(report) {
 }
 
 // ../../../packages/cli/src/main.ts
-var VERSION = "0.7.9";
+var VERSION = "0.7.12";
 async function main(argv) {
   const parsed = parseArgs(argv);
   switch (parsed.command) {