@reconcrap/boss-recommend-mcp 2.0.57 → 2.1.1

package/README.md

@@ -20,6 +20,7 @@ Boss 推荐 / 搜索 / 聊天筛选 MCP（stdio）服务。
 MCP 工具：
 
 - `list_recommend_jobs`（只读读取推荐页岗位下拉框，返回可直接用于 cron/一次性任务的 `job_names`）
+- `prepare_recommend_pipeline_run`（只校验完整 payload 是否已可用于 cron/一次性任务；不启动筛选，返回 `READY + cron_ready=true` 后才应创建 cron）
 - `start_recommend_pipeline_run`（异步启动；同样先经过前置门禁，通过后返回 run_id）
 - `get_recommend_pipeline_run`（轮询 run_id 状态）
 - `cancel_recommend_pipeline_run`（取消运行中任务）
@@ -53,10 +54,19 @@ boss-recommend-mcp list-jobs --slow-live --port 9222
 
 返回的 `job_names` 可直接作为后续 `start_recommend_pipeline_run` 的 `confirmation.job_value` / `overrides.job`。
 
+Cron / 一次性定时任务设置建议先在设置阶段完成 Chrome/登录/岗位发现与全部确认：
+
+```bash
+boss-recommend-mcp prepare-run --instruction-file boss-recommend-instruction.txt --overrides-file boss-recommend-overrides.json --confirmation-file boss-recommend-confirmation.json --slow-live --port 9222
+```
+
+只有输出 `status: "READY"` 且 `cron_ready: true` 后，才把同一份 payload 写入 cron 并在到点时调用 `start_recommend_pipeline_run` / `boss-recommend-mcp run --detached`。如果设置阶段返回 `BOSS_LOGIN_REQUIRED`、`NEED_INPUT` 或 `NEED_CONFIRMATION`，先让用户登录或补齐确认，不要创建 cron。
+
 状态机：
 
 - `NEED_INPUT`
 - `NEED_CONFIRMATION`
+- `READY`（仅准备工具）
 - `COMPLETED`
 - `FAILED`
 
@@ -259,13 +269,16 @@ config/screening-config.example.json
 - `debugPort`：未显式传 `port` 时，recommend / search / chat CDP-only MCP run 和健康检查默认连接这个 Chrome 调试端口。
 - `outputDir`：recommend / search / chat 完成后的最终 CSV 与 report JSON 会写入这里；run state / checkpoint 仍保留在各自状态目录，方便 pause/resume/cancel。
 - `llmThinkingLevel`：默认 `low`。可设为 `off/minimal/low/medium/high/auto/current`，用于控制 OpenAI-compatible LLM 的 thinking/reasoning 强度。
-- `humanBehavior`：默认 `{ "enabled": true, "profile": "paced_with_rests" }`。用于 recommend / search / chat 的可靠性实验，支持：
+- `humanBehavior`：默认 `{ "enabled": true, "profile": "paced_with_rests", "restLevel": "low" }`。用于 recommend / search / chat 的可靠性实验，支持：
   - `profile: "baseline"`：关闭人类节奏，保持确定性行为。
   - `profile: "paced"`：启用 CDP-only Bezier 鼠标移动、较大按钮的安全 inset 点击点、分块 `Input.insertText`、列表 wheel/settle jitter，以及小的动作前后读秒。
   - `profile: "paced_with_rests"`：在 `paced` 基础上启用候选人短休和批次休息。
+  - `restLevel: "low"`：保持旧版休息策略不变，候选人短休 8% 概率暂停 3-7 秒，批次休息约每 25-32 人暂停 15-30 秒。
+  - `restLevel: "medium"`：随机分散短/长休息，平均目标约每 5 小时或 700 位候选人累计休息 30 分钟。
+  - `restLevel: "high"`：随机分散短/长休息，平均目标约每 5 小时或 700 位候选人累计休息 1 小时。
 - `humanRestEnabled`：兼容旧配置。设为 `true` 时等价于 `humanBehavior.profile="paced_with_rests"`；设为 `false` 时不会关闭当前默认节奏。如需关闭，请显式设置 `humanBehavior.enabled=false` 或 `humanBehavior.profile="baseline"`。
   - recommend / search / chat 图片简历 fallback 与主列表滚动都会在启用 `listScrollJitter` 时使用 coverage-safe scroll jitter：每次 wheel delta 在安全范围内变化，并保留截图重叠、重复检测、bottom-marker / stop-boundary 逻辑，实际 delta 和 settle 时间会写入 artifact metadata。
-  - chat/recommend/search run 也兼容显式参数 `safe_pacing` 与 `batch_rest_enabled`：run 参数优先于配置文件。
+  - chat/recommend/search run 也兼容显式参数 `safe_pacing`、`batch_rest_enabled` 与 `human_behavior.restLevel`：run 参数优先于配置文件。AI harness/skill 启动每次 run 前必须让用户明确选择 `low/medium/high`，再把选择写入 `human_behavior.restLevel`。
 
 ## 常用命令
 
@@ -371,6 +384,7 @@ Trae-CN / 长对话防循环建议：
 说明：
 
 - `start_recommend_pipeline_run` 为异步入口，但不会跳过同步确认流程。
+- `prepare_recommend_pipeline_run` / `boss-recommend-mcp prepare-run` 用于 cron 设置阶段；它不启动筛选，只确认 payload 已经不需要到点后再问用户。
 - 定时心跳默认 120 秒一次；`updated_at` 仍会在阶段或进度变化时刷新。
 - 每个 run 会持久化到 `~/.boss-recommend-mcp/runs/<run_id>.json`（可通过 `BOSS_RECOMMEND_HOME` 覆盖）。
 - screen 阶段会持久化 checkpoint：`~/.boss-recommend-mcp/runs/<run_id>.checkpoint.json`。

package/config/screening-config.example.json

@@ -16,6 +16,7 @@
   "humanBehavior": {
     "enabled": true,
     "profile": "paced_with_rests",
+    "restLevel": "low",
     "clickMovement": true,
     "textEntry": true,
     "listScrollJitter": true,

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@reconcrap/boss-recommend-mcp",
-  "version": "2.0.57",
+  "version": "2.1.1",
   "description": "Unified MCP pipeline for recommend-page filtering and screening on Boss Zhipin",
   "keywords": [
     "boss",

package/skills/boss-chat/README.md

@@ -17,7 +17,7 @@ Please run a Boss chat-only task (do not switch to recommend flow).
 Execution order:
 1) Call boss_chat_health_check.
 2) Call prepare_boss_chat_run once (empty params allowed) to fetch job_options and missing fields.
-3) Ask for these required fields in one shot: job, start_from (unread/all), target_count, criteria.
+3) Ask for these required fields in one shot: job, start_from (unread/all), target_count, criteria, rest_level (low/medium/high).
 4) After user reply, call start_boss_chat_run exactly once to start the run.
 5) If ACCEPTED, reply only with run_id and "task started"; no auto polling.
 
@@ -27,6 +27,7 @@ Anti-loop rules:
 - Do not use start_boss_chat_run for preflight. It is only for the final start call and must include job/start_from/target_count/criteria.
 - Do not call start_boss_chat_run repeatedly in one turn.
 - Do not call get_boss_chat_run unless user explicitly asks for progress.
+- Do not choose rest_level for the user. Pass the user's choice as `human_behavior.restLevel`.
 
 target_count mapping:
 - Positive integer means explicit cap (for example 20).

package/skills/boss-chat/SKILL.md

@@ -27,6 +27,7 @@ description: "Use when users want Boss chat-page screening/outreach via the bund
 - `start_from`: `unread|all`
 - `target_count`
 - `criteria`
+- `rest_level`: `low|medium|high`
 
 可选：
 
@@ -38,6 +39,12 @@ description: "Use when users want Boss chat-page screening/outreach via the bund
 - `safe_pacing`
 - `batch_rest_enabled`
 
+启动工具时，把用户确认的休息强度写入 `human_behavior.restLevel`，例如：
+
+```json
+{ "human_behavior": { "restLevel": "medium" } }
+```
+
 `greeting_text` 默认规则：
 
 - 本次显式传入 `greeting_text`：使用本次值
@@ -62,6 +69,7 @@ description: "Use when users want Boss chat-page screening/outreach via the bund
 - 若本机找不到 Chrome，可提示用户设置 `BOSS_MCP_CHROME_PATH` 或 `BOSS_RECOMMEND_CHROME_PATH`；非本机 debug host 不自动启动。
 - `job` / `start_from` / `criteria` 缺一不可；缺参时只补缺口。
 - `target_count` 在 chat-only 启动前也是必填项，不能默认省略。
+- 每次 run 必须明确询问用户本次休息强度 `rest_level`：`low`（旧策略）/ `medium`（约 5 小时或 700 人累计休息 30 分钟）/ `high`（约 5 小时或 700 人累计休息 1 小时）；不得默认使用配置文件里的值替用户决定。
 - 当用户说“全部候选人/所有候选人”时，必须按“扫到底（unlimited）”处理，不要再追问正整数。
 - 参数名必须写 `target_count`（不要写“目标数量”等中文键名）。
 - 当用户选择“扫到底/全部候选人/所有候选人”时，调用参数优先写：`"target_count": "all"`；`-1` 只作为兼容输入和内部 CLI 表示。
@@ -89,5 +97,5 @@ description: "Use when users want Boss chat-page screening/outreach via the bund
 
 - 用结构化中文。
 - 首轮建议先调用一次 `prepare_boss_chat_run`（可空参）获取 `job_options` 与 `pending_questions`。
-- 缺参时必须逐项确认：`job`（来自岗位列表）、`start_from`（`unread|all`）、`target_count`、`criteria`。
+- 缺参时必须逐项确认：`job`（来自岗位列表）、`start_from`（`unread|all`）、`target_count`、`criteria`、`rest_level`。
 - 若健康检查失败，明确提示共享配置文件 `screening-config.json` 不可用。

package/skills/boss-recommend-pipeline/README.md

@@ -7,6 +7,7 @@
 - 先确认推荐页 filters
 - 再确认筛选 criteria
 - 再确认本次运行统一动作 `favorite` 或 `greet`
+- 每次运行都要让用户明确选择休息强度 `low` / `medium` / `high`，并传入 `human_behavior.restLevel`
 - 只确认一次 `post_action`，不要逐个候选人反复确认
 - 运行中临时中断请使用 `pause_recommend_pipeline_run`（按 run_id），不要靠自然语言“暂停/继续”指令
 - 继续执行请使用 `resume_recommend_pipeline_run`；状态查询默认按用户指令触发，不自动轮询

package/skills/boss-recommend-pipeline/SKILL.md

@@ -20,7 +20,8 @@ description: "Use when users want Boss recommend-page filtering/screening via bo
 - **确认不可代填（强制）**
   - 禁止 agent 自行“设置合理参数”并代替用户确认。
   - 禁止在用户未明确回复前，把任意 `*_confirmed` 字段设为 `true`。
-  - 禁止在用户未明确回复前，自行填充 `page_scope/school_tag/degree/gender/recent_not_view/criteria/target_count/post_action/max_greet_count/job`。
+  - 禁止在用户未明确回复前，自行填充 `page_scope/school_tag/degree/gender/recent_not_view/criteria/target_count/post_action/max_greet_count/job/rest_level`。
+  - 每次 run 必须明确询问用户本次休息强度 `rest_level`：`low`（旧策略）/ `medium`（约 5 小时或 700 人累计休息 30 分钟）/ `high`（约 5 小时或 700 人累计休息 1 小时）；不得默认使用配置文件里的值替用户决定。
   - 若工具返回 `pending_questions`，必须逐项向用户提问并等待用户回复；不得跳过提问直接执行。
 
 - **岗位确认时机**
@@ -54,6 +55,7 @@ description: "Use when users want Boss recommend-page filtering/screening via bo
 - `target_count`（可空）
 - `post_action`：`favorite|greet|none`
 - `max_greet_count`（仅当 `post_action=greet`）
+- `rest_level`：`low|medium|high`
 
 ### Stage B (页面就绪后)
 
@@ -82,18 +84,24 @@ description: "Use when users want Boss recommend-page filtering/screening via bo
   - `gender`: `不限/男/女`
   - `recent_not_view`: `不限/近14天没有`
   - `post_action`: `favorite/greet/none`
+  - `rest_level`: `low/medium/high`
 
 ## Tool Usage
 
 - 岗位发现工具：`list_recommend_jobs`
-  - 用途：当用户需要为 cron / 一次性自动任务提前填写完整参数时，先用它读取推荐页岗位下拉框的全部可用岗位名。
+  - 用途：当用户需要为 cron / 一次性自动任务提前填写完整参数时，先用它读取推荐页岗位下拉框的全部可用岗位名；默认会复用/自动打开本机 9222 Chrome 并导航到推荐页。
   - 输出：优先把 `job_names` 里的值作为后续 `overrides.job` / `confirmation.job_value`。
-  - 限制：只读岗位列表，不启动筛选任务；仍然必须在正式 `start_recommend_pipeline_run` 前完成岗位与最终确认。
+  - 限制：只读岗位列表，不启动筛选任务；若返回 `BOSS_LOGIN_REQUIRED`，必须让用户在自动打开的 Chrome 完成登录后重试，本次 cron 不得创建。
+- Cron 准备工具：`prepare_recommend_pipeline_run`
+  - 用途：只校验参数是否已可用于 cron / 一次性任务，不启动筛选任务。
+  - 要求：只有返回 `status=READY` 且 `cron_ready=true` 后，才允许创建 cron。
+  - 若返回 `NEED_INPUT` / `NEED_CONFIRMATION` / `FAILED`：继续补齐 `pending_questions` 或修复登录/页面/config；不得先创建 cron。
 - 主工具：`start_recommend_pipeline_run`
 - 必填：`instruction`
 - 关键输入：
   - `confirmation`：`page_confirmed/page_value/filters_confirmed/school_tag_confirmed.../job_confirmed/job_value/final_confirmed`
   - `overrides`：`page_scope/school_tag/degree/gender/recent_not_view/criteria/job/target_count/post_action/max_greet_count`
+  - `human_behavior`：必须包含本次用户确认的 `restLevel`（例如 `{ "restLevel": "medium" }`）
   - 不要传 `follow_up.chat`；该路径属于 legacy-only 行为
 
 最小策略：
@@ -103,6 +111,25 @@ description: "Use when users want Boss recommend-page filtering/screening via bo
 - 拿到 `ACCEPTED + run_id` 后默认停止本轮，不自动轮询。
 - 在拿到 `ACCEPTED + run_id` 之前，禁止以“我已帮你确认参数”为由越过必填确认阶段。
 
+## Cron / Scheduled Run Policy
+
+创建 cron 时必须在设置 cron 的当下完成全部交互：
+
+1. 锁定用户原始 instruction，不改写、不摘要，criteria 放入 `overrides.criteria` 时必须逐字保留。
+2. 收集 Stage A 全部筛选项、`target_count`、`post_action`、`max_greet_count`（如需要）和本次 `rest_level`。
+3. 调用 `list_recommend_jobs`；若 Chrome 未打开，工具会尝试自动打开本机 9222 Chrome 并进入推荐页。若返回 `BOSS_LOGIN_REQUIRED` 或页面不可用，停止 cron 创建，让用户登录/修复后重试。
+4. 用 `job_names` 中的精确岗位名完成岗位确认，并完成最终总确认，写入 `job_confirmed=true` 与 `final_confirmed=true`。
+5. 调用 `prepare_recommend_pipeline_run` 传入将来要执行的完整 payload；只有 `READY + cron_ready=true` 才能创建 cron。
+6. cron 到点时只调用 `start_recommend_pipeline_run`，并传入准备阶段验证过的同一份 payload。到点后若 Chrome/登录异常，应作为运行失败处理，不得再向用户追问参数。
+
+Shell-only OpenClaw/QClaw cron 设置同样先运行：
+
+```powershell
+npx -y @reconcrap/boss-recommend-mcp@latest prepare-run --instruction-file .\boss-recommend-instruction.txt --overrides-file .\boss-recommend-overrides.json --confirmation-file .\boss-recommend-confirmation.json --slow-live --port 9222
+```
+
+仅当上述命令输出 `READY` 且 `cron_ready=true` 后，才把对应的 detached `run` 命令写入 cron。
+
 ## Async Run Policy
 
 - 用户未明确要求“持续跟进”时，不自动 `sleep + get_recommend_pipeline_run`。
@@ -146,13 +173,14 @@ description: "Use when users want Boss recommend-page filtering/screening via bo
 推荐做法：
 
 1. 将锁定的用户原文写入 instruction 文件，将已确认参数写入 `overrides` 与 `confirmation` JSON 文件。
-2. 用 detached CLI 启动，让父命令返回启动证据，子进程继续持有 CDP 会话：
+2. 先用 `prepare-run` 校验完整 payload 已 cron-ready；未返回 `READY + cron_ready=true` 不得创建 cron 或启动 run。
+3. 用 detached CLI 启动，让父命令返回启动证据，子进程继续持有 CDP 会话：
 
 ```powershell
 npx -y @reconcrap/boss-recommend-mcp@latest run --detached --instruction-file .\boss-recommend-instruction.txt --overrides-file .\boss-recommend-overrides.json --confirmation-file .\boss-recommend-confirmation.json --slow-live --port 9222
 ```
 
-3. 若返回 `ACCEPTED + run_id`，即任务已启动；记录 `run_id/stdout_path/stderr_path`。若返回 `NEED_INPUT` 或 `NEED_CONFIRMATION`，只补 `pending_questions`，不要重写已锁定的用户原文。
+4. 若返回 `ACCEPTED + run_id`，即任务已启动；记录 `run_id/stdout_path/stderr_path`。若返回 `NEED_INPUT` 或 `NEED_CONFIRMATION`，说明 cron 设置阶段漏掉了准备门禁，应回到 `prepare-run` 补齐，不能在到点后继续问用户确认。
 
 兼容路径：
 

package/skills/boss-recruit-pipeline/README.md

@@ -15,3 +15,5 @@ This skill intentionally replaces legacy `boss-recruit-mcp` skill installs. It r
 - `cancel_recruit_pipeline_run`
 
 Do not call the old `@reconcrap/boss-recruit-mcp` package from this skill.
+
+Each run must ask the user to choose `rest_level` (`low` / `medium` / `high`) and pass the answer as `human_behavior.restLevel`; do not pick a default for the user.

package/skills/boss-recruit-pipeline/SKILL.md

@@ -32,6 +32,7 @@ description: "Use when users want Boss search/recruit-page screening via the uni
 - 若用户提供城市、学历、学校、关键词、过滤已看、人选目标数、筛选条件、post action、max greet 等参数，必须逐项传入或确认。
 - `post_action=greet` 时必须确认 `max_greet_count`；不要默认等于 `target_count`。
 - 搜索页和推荐页一样支持多选筛选条件；不要把多选降级成单选。
+- 每次 run 必须明确询问用户本次休息强度 `rest_level`：`low`（旧策略）/ `medium`（约 5 小时或 700 人累计休息 30 分钟）/ `high`（约 5 小时或 700 人累计休息 1 小时）；不得默认使用配置文件里的值替用户决定。
 
 ## Required Inputs
 
@@ -39,6 +40,7 @@ description: "Use when users want Boss search/recruit-page screening via the uni
 - `keyword` 或用户明确的搜索意图
 - `criteria`
 - `target_count`
+- `rest_level`: `low|medium|high`
 
 常用可选项：
 
@@ -50,6 +52,12 @@ description: "Use when users want Boss search/recruit-page screening via the uni
 - `max_greet_count`
 - `port`
 
+启动工具时，把用户确认的休息强度写入 `human_behavior.restLevel`，例如：
+
+```json
+{ "human_behavior": { "restLevel": "medium" } }
+```
+
 ## Response Style
 
 - 用结构化中文确认参数。

package/src/cli.js

@@ -23,10 +23,11 @@ import {
   prepareBossChatRunTool,
   resumeBossChatRunTool
 } from "./chat-mcp.js";
-import {
-  listRecommendJobsTool,
-  startRecommendPipelineRunTool
-} from "./recommend-mcp.js";
+import {
+  listRecommendJobsTool,
+  prepareRecommendPipelineRunTool,
+  startRecommendPipelineRunTool
+} from "./recommend-mcp.js";
 import {
   getBossScreenConfigResolution,
   resolveBossChatRuntimeLayout as resolveCdpBossChatRuntimeLayout,
@@ -53,7 +54,7 @@ const defaultMcpServerName = "boss-recommend";
 const defaultMcpCommand = "npx";
 const recommendMcpPackageName = "@reconcrap/boss-recommend-mcp";
 const recommendMcpBinaryName = "boss-recommend-mcp";
-const autoSyncSkipCommands = new Set(["install", "install-skill", "where", "help", "--help", "-h", "list-jobs", "jobs", "recommend-jobs"]);
+const autoSyncSkipCommands = new Set(["install", "install-skill", "where", "help", "--help", "-h", "list-jobs", "jobs", "recommend-jobs"]);
 const externalMcpTargetsEnv = "BOSS_RECOMMEND_MCP_CONFIG_TARGETS";
 const externalSkillDirsEnv = "BOSS_RECOMMEND_EXTERNAL_SKILL_DIRS";
 const installConfigDefaults = Object.freeze({
@@ -63,12 +64,13 @@ const installConfigDefaults = Object.freeze({
   llmMaxRetries: 3,
   llmImageLimit: 8,
   llmImageDetail: "low",
-  humanRestEnabled: true,
-  humanBehavior: {
-    enabled: true,
-    profile: "paced_with_rests"
-  }
-});
+  humanRestEnabled: true,
+  humanBehavior: {
+    enabled: true,
+    profile: "paced_with_rests",
+    restLevel: "low"
+  }
+});
 const bossChatRuntimeChildDirs = ["logs", "runs", "profiles", "reports", "artifacts", "state"];
 const bossChatCliUnsupportedStartCode = "CHAT_CLI_ASYNC_UNSUPPORTED_CDP_ONLY";
 const calibrateUnsupportedCode = "CALIBRATE_UNSUPPORTED_CDP_ONLY";
@@ -613,14 +615,20 @@ function parseBossChatTargetCountOption(raw) {
   return parsed ?? text;
 }
 
-function parseBooleanOption(raw, fallback = undefined) {
-  if (raw === undefined || raw === null || raw === "") return fallback;
-  if (raw === true) return true;
-  const normalized = String(raw).trim().toLowerCase();
-  if (["true", "1", "yes", "y", "on"].includes(normalized)) return true;
-  if (["false", "0", "no", "n", "off"].includes(normalized)) return false;
-  return fallback;
-}
+function parseBooleanOption(raw, fallback = undefined) {
+  if (raw === undefined || raw === null || raw === "") return fallback;
+  if (raw === true) return true;
+  const normalized = String(raw).trim().toLowerCase();
+  if (["true", "1", "yes", "y", "on"].includes(normalized)) return true;
+  if (["false", "0", "no", "n", "off"].includes(normalized)) return false;
+  return fallback;
+}
+
+function parseRestLevelOption(raw) {
+  if (raw === undefined || raw === null || raw === "") return undefined;
+  const normalized = String(raw).trim().toLowerCase();
+  return ["low", "medium", "high"].includes(normalized) ? normalized : undefined;
+}
 
 function normalizePageScope(value) {
   const normalized = String(value || "").trim().toLowerCase();
@@ -1271,20 +1279,33 @@ async function resolveCliConfigTarget(options = {}) {
   };
 }
 
-function applyMissingInstallConfigDefaults(config = {}) {
-  const nextConfig = { ...config };
-  const patchedKeys = [];
-  for (const [key, defaultValue] of Object.entries(installConfigDefaults)) {
+function applyMissingInstallConfigDefaults(config = {}) {
+  const nextConfig = { ...config };
+  const patchedKeys = [];
+  for (const [key, defaultValue] of Object.entries(installConfigDefaults)) {
     if (!Object.prototype.hasOwnProperty.call(nextConfig, key)) {
       nextConfig[key] = defaultValue;
-      patchedKeys.push(key);
-    }
-  }
-  return {
-    nextConfig,
-    patchedKeys
-  };
-}
+      patchedKeys.push(key);
+    }
+  }
+  if (
+    nextConfig.humanBehavior
+    && typeof nextConfig.humanBehavior === "object"
+    && !Array.isArray(nextConfig.humanBehavior)
+    && !Object.prototype.hasOwnProperty.call(nextConfig.humanBehavior, "restLevel")
+    && !Object.prototype.hasOwnProperty.call(nextConfig.humanBehavior, "rest_level")
+  ) {
+    nextConfig.humanBehavior = {
+      ...nextConfig.humanBehavior,
+      restLevel: "low"
+    };
+    patchedKeys.push("humanBehavior.restLevel");
+  }
+  return {
+    nextConfig,
+    patchedKeys
+  };
+}
 
 async function ensureUserConfig(options = {}) {
   const { configPath, workspacePreferred } = await resolveCliConfigTarget(options);
@@ -2601,10 +2622,11 @@ function printHelp() {
   console.log("boss-recommend-mcp");
   console.log("");
   console.log("Usage:");
-  console.log("  boss-recommend-mcp              Start the MCP server");
-  console.log("  boss-recommend-mcp start        Start the MCP server");
-  console.log("  boss-recommend-mcp run          Start a CDP-only recommend run through the shared run service");
-  console.log("  boss-recommend-mcp list-jobs    CDP-only list of exact recommend job names for cron/one-shot inputs");
+  console.log("  boss-recommend-mcp              Start the MCP server");
+  console.log("  boss-recommend-mcp start        Start the MCP server");
+  console.log("  boss-recommend-mcp prepare-run  Validate a cron-ready recommend run payload without starting screening");
+  console.log("  boss-recommend-mcp run          Start a CDP-only recommend run through the shared run service");
+  console.log("  boss-recommend-mcp list-jobs    CDP-only list of exact recommend job names for cron/one-shot inputs");
   console.log("  boss-recommend-mcp chat <subcommand>  Run CDP-only boss-chat health/prepare/status commands");
   console.log("  boss-recommend-mcp install      Install/migrate skills and MCP configs; replaces legacy Boss MCP routes (supports --agent trae-cn/openclaw/qclaw/...)");
   console.log("  boss-recommend-mcp install-skill Install bundled Codex skills (recommend/recruit/chat)");
@@ -2616,10 +2638,11 @@ function printHelp() {
   console.log("  boss-recommend-mcp calibrate    Disabled until CDP-only featured calibration is live-verified");
   console.log("  boss-recommend-mcp launch-chrome Launch or reuse Chrome debug instance and open Boss recommend page");
   console.log("  boss-recommend-mcp where        Print installed package, skill, and config paths");
-  console.log("");
-  console.log("Run command:");
-  console.log("  boss-recommend-mcp run --instruction \"推荐页上筛选211男生，近14天没有，有大模型平台经验\" --overrides-file overrides.json --confirmation-file confirmation.json");
-  console.log("  boss-recommend-mcp run --detached --instruction \"...\" --overrides-file overrides.json --confirmation-file confirmation.json");
+  console.log("");
+  console.log("Run command:");
+  console.log("  boss-recommend-mcp prepare-run --instruction \"...\" --overrides-file overrides.json --confirmation-file confirmation.json");
+  console.log("  boss-recommend-mcp run --instruction \"推荐页上筛选211男生，近14天没有，有大模型平台经验\" --overrides-file overrides.json --confirmation-file confirmation.json");
+  console.log("  boss-recommend-mcp run --detached --instruction \"...\" --overrides-file overrides.json --confirmation-file confirmation.json");
   console.log("  boss-recommend-mcp list-jobs --slow-live --port 9222");
   console.log("  boss-recommend-mcp chat prepare-run --slow-live --port 9222    # CDP-only preflight; start runs through MCP start_boss_chat_run");
   console.log("  boss-recommend-mcp config set --base-url <url> --api-key <key> --model <model> [--thinking-level off|low|medium|high|current] [--greeting-message <text>] [--openai-organization <id>] [--openai-project <id>]");
@@ -2713,27 +2736,37 @@ async function installAll(options = {}) {
   }
 }
 
-async function runPipelineOnce(options = {}) {
-  const instruction = getRunInstruction(options);
-  const confirmation = getRunConfirmation(options);
-  const overrides = getRunOverrides(options);
-  const followUp = getRunFollowUp(options);
-  const workspaceRoot = getWorkspaceRoot(options);
-  const port = parsePositivePort(options.port) || parsePositivePort(process.env.BOSS_RECOMMEND_CHROME_PORT) || 9222;
-
-  const args = {
-    instruction,
-    confirmation: confirmation ?? undefined,
-    overrides: overrides ?? undefined,
-    follow_up: followUp ?? undefined,
+function buildRecommendRunCliInput(options = {}) {
+  const instruction = getRunInstruction(options);
+  const confirmation = getRunConfirmation(options);
+  const overrides = getRunOverrides(options);
+  const followUp = getRunFollowUp(options);
+  const port = parsePositivePort(options.port) || parsePositivePort(process.env.BOSS_RECOMMEND_CHROME_PORT) || 9222;
+
+  const args = {
+    instruction,
+    confirmation: confirmation ?? undefined,
+    overrides: overrides ?? undefined,
+    follow_up: followUp ?? undefined,
     host: typeof options.host === "string" && options.host.trim() ? options.host.trim() : undefined,
     port,
     target_url_includes: typeof options["target-url-includes"] === "string" && options["target-url-includes"].trim()
       ? options["target-url-includes"].trim()
       : undefined,
     allow_navigate: !(options["no-navigate"] === true || options.noNavigate === true || options.allow_navigate === false),
-    slow_live: options["slow-live"] === true || options.slowLive === true || options.slow_live === true
-  };
+    slow_live: options["slow-live"] === true || options.slowLive === true || options.slow_live === true
+  };
+  const restLevel = parseRestLevelOption(
+    options["rest-level"]
+    ?? options.rest_level
+    ?? options["human-behavior-rest-level"]
+    ?? options.human_behavior_rest_level
+  );
+  if (restLevel) {
+    args.human_behavior = {
+      restLevel
+    };
+  }
 
   const optionalPassthrough = [
     "detail_limit",
@@ -2766,15 +2799,46 @@ async function runPipelineOnce(options = {}) {
     "llm_image_limit",
     "llm_image_detail"
   ];
-  for (const key of optionalPassthrough) {
-    const kebab = key.replace(/_/g, "-");
-    if (options[key] !== undefined) args[key] = options[key];
-    else if (options[kebab] !== undefined) args[key] = options[kebab];
-  }
-
-  const result = await startRecommendPipelineRunTool({
-    workspaceRoot,
-    args
+  for (const key of optionalPassthrough) {
+    const kebab = key.replace(/_/g, "-");
+    if (options[key] !== undefined) args[key] = options[key];
+    else if (options[kebab] !== undefined) args[key] = options[kebab];
+  }
+
+  return {
+    args,
+    port
+  };
+}
+
+async function preparePipelineOnce(options = {}) {
+  const workspaceRoot = getWorkspaceRoot(options);
+  const { args, port } = buildRecommendRunCliInput(options);
+  const result = prepareRecommendPipelineRunTool({
+    workspaceRoot,
+    args
+  });
+  printJson({
+    ...result,
+    cli: {
+      command: "prepare-run",
+      cdp_only: true,
+      shared_run_service: true,
+      workspace_root: workspaceRoot,
+      port
+    }
+  });
+  if (result.status !== "READY" || result.cron_ready !== true) {
+    process.exitCode = 1;
+  }
+}
+
+async function runPipelineOnce(options = {}) {
+  const workspaceRoot = getWorkspaceRoot(options);
+  const { args, port } = buildRecommendRunCliInput(options);
+  const result = await startRecommendPipelineRunTool({
+    workspaceRoot,
+    args
   });
   printJson({
     ...result,
@@ -2810,16 +2874,22 @@ async function listRecommendJobsCli(options = {}) {
   }));
 }
 
-function buildBossChatCliInput(options = {}) {
-  const greetingTextRaw =
-    options["greeting-text"]
-    ?? options.greeting_text
-    ?? options.greetingText
-    ?? options.greeting;
-  const greetingText = typeof greetingTextRaw === "string" ? greetingTextRaw.trim() : undefined;
-  const targetUrlIncludes = String(options["target-url-includes"] || options.target_url_includes || "").trim();
-  const host = String(options.host || "").trim();
-  return {
+function buildBossChatCliInput(options = {}) {
+  const greetingTextRaw =
+    options["greeting-text"]
+    ?? options.greeting_text
+    ?? options.greetingText
+    ?? options.greeting;
+  const greetingText = typeof greetingTextRaw === "string" ? greetingTextRaw.trim() : undefined;
+  const targetUrlIncludes = String(options["target-url-includes"] || options.target_url_includes || "").trim();
+  const host = String(options.host || "").trim();
+  const restLevel = parseRestLevelOption(
+    options["rest-level"]
+    ?? options.rest_level
+    ?? options["human-behavior-rest-level"]
+    ?? options.human_behavior_rest_level
+  );
+  return {
     profile: typeof options.profile === "string" ? options.profile.trim() : undefined,
     job: typeof options.job === "string" ? options.job.trim() : undefined,
     start_from: String(options["start-from"] || options.start_from || "").trim().toLowerCase() || undefined,
@@ -2837,13 +2907,18 @@ function buildBossChatCliInput(options = {}) {
     dry_run: options["dry-run"] === true || options.dryRun === true,
     no_state: options["no-state"] === true || options.noState === true,
     human_behavior_enabled: parseBooleanOption(options["human-behavior-enabled"] ?? options.human_behavior_enabled),
-    human_behavior_profile: typeof (options["human-behavior-profile"] ?? options.human_behavior_profile) === "string"
-      ? (options["human-behavior-profile"] ?? options.human_behavior_profile).trim()
-      : undefined,
-    safe_pacing: parseBooleanOption(options["safe-pacing"] ?? options.safe_pacing),
-    batch_rest_enabled: parseBooleanOption(options["batch-rest"] ?? options.batch_rest_enabled)
-  };
-}
+    human_behavior_profile: typeof (options["human-behavior-profile"] ?? options.human_behavior_profile) === "string"
+      ? (options["human-behavior-profile"] ?? options.human_behavior_profile).trim()
+      : undefined,
+    human_behavior: restLevel
+      ? {
+          restLevel
+        }
+      : undefined,
+    safe_pacing: parseBooleanOption(options["safe-pacing"] ?? options.safe_pacing),
+    batch_rest_enabled: parseBooleanOption(options["batch-rest"] ?? options.batch_rest_enabled)
+  };
+}
 
 function getBossChatCliRunTarget(options = {}) {
   return {
@@ -2932,11 +3007,11 @@ export async function runCli(argv = process.argv) {
     case "mcp":
       startServer();
       break;
-    case "run":
-      try {
-        if (
-          (options.detached === true || options.background === true)
-          && process.env[detachedRecommendRunChildEnv] !== "1"
+    case "run":
+      try {
+        if (
+          (options.detached === true || options.background === true)
+          && process.env[detachedRecommendRunChildEnv] !== "1"
         ) {
           await runPipelineDetached(argv.slice(3), options);
         } else {
@@ -2951,9 +3026,26 @@ export async function runCli(argv = process.argv) {
             retryable: false
           }
         });
-        process.exitCode = 1;
-      }
-      break;
+        process.exitCode = 1;
+      }
+      break;
+    case "prepare-run":
+    case "prepare":
+      try {
+        await preparePipelineOnce(options);
+      } catch (error) {
+        printJson({
+          status: "FAILED",
+          cron_ready: false,
+          error: {
+            code: "INVALID_CLI_INPUT",
+            message: error.message || "Invalid CLI input",
+            retryable: false
+          }
+        });
+        process.exitCode = 1;
+      }
+      break;
     case "list-jobs":
     case "jobs":
     case "recommend-jobs":
@@ -3111,10 +3203,11 @@ export const __testables = {
   installSkill,
   isInstalledPackageRoot,
   mergeMcpServerConfigFile,
-  resolveBossChatRuntimeLayout: resolveCdpBossChatRuntimeLayout,
-  runBossChatCliCommand,
-  runPipelineOnce
-};
+  resolveBossChatRuntimeLayout: resolveCdpBossChatRuntimeLayout,
+  runBossChatCliCommand,
+  preparePipelineOnce,
+  runPipelineOnce
+};
 
 if (process.argv[1] && path.resolve(process.argv[1]) === currentFilePath) {
   await runCli(process.argv);

package/src/core/browser/index.js

@@ -86,6 +86,46 @@ const HUMAN_BEHAVIOR_PROFILE_ALIASES = Object.freeze({
   rests: "paced_with_rests",
   rest: "paced_with_rests"
 });
+const DEFAULT_HUMAN_REST_LEVEL = "low";
+const HUMAN_REST_LEVEL_ALIASES = Object.freeze({
+  default: "low",
+  light: "low",
+  normal: "medium",
+  med: "medium",
+  heavy: "high"
+});
+const HUMAN_REST_LEVEL_PROFILES = Object.freeze({
+  medium: Object.freeze({
+    targetRestMs: 30 * 60 * 1000,
+    targetCandidateCount: 700,
+    targetWindowMs: 5 * 60 * 60 * 1000,
+    intervalMin: 4,
+    intervalMax: 16,
+    longRestProbability: 0.22,
+    shortRestMinMs: 8000,
+    shortRestMaxMs: 45000,
+    longRestMinMs: 60000,
+    longRestMaxMs: 180000,
+    minDebtToRestMs: 8000,
+    forceDebtMs: 90000,
+    maxOverspendMs: 15000
+  }),
+  high: Object.freeze({
+    targetRestMs: 60 * 60 * 1000,
+    targetCandidateCount: 700,
+    targetWindowMs: 5 * 60 * 60 * 1000,
+    intervalMin: 3,
+    intervalMax: 12,
+    longRestProbability: 0.28,
+    shortRestMinMs: 12000,
+    shortRestMaxMs: 75000,
+    longRestMinMs: 90000,
+    longRestMaxMs: 300000,
+    minDebtToRestMs: 12000,
+    forceDebtMs: 150000,
+    maxOverspendMs: 25000
+  })
+});
 
 function clampNumber(value, min, max) {
   const number = Number(value);
@@ -152,6 +192,14 @@ export function normalizeHumanBehaviorProfile(raw, fallback = "baseline") {
     : fallback;
 }
 
+export function normalizeHumanRestLevel(raw, fallback = DEFAULT_HUMAN_REST_LEVEL) {
+  const normalized = String(raw || "").trim().toLowerCase().replace(/[\s-]+/g, "_");
+  const level = HUMAN_REST_LEVEL_ALIASES[normalized] || normalized;
+  return level === "low" || level === "medium" || level === "high"
+    ? level
+    : fallback;
+}
+
 export function normalizeHumanBehaviorOptions(raw = null, {
   legacyEnabled = false,
   safePacing = null,
@@ -231,6 +279,10 @@ export function normalizeHumanBehaviorOptions(raw = null, {
     readFirstOption(rawObject, ["batchRest", "batch_rest", "batchRestEnabled", "batch_rest_enabled"]),
     profileDefaults.batchRest
   );
+  const restLevel = normalizeHumanRestLevel(
+    readFirstOption(rawObject, ["restLevel", "rest_level"]),
+    DEFAULT_HUMAN_REST_LEVEL
+  );
   if (batchRestFlag !== null) {
     batchRest = batchRestFlag;
     if (batchRestFlag === true && readFirstOption(rawObject, ["shortRest", "short_rest", "randomRest", "random_rest"]) === undefined) {
@@ -248,6 +300,7 @@ export function normalizeHumanBehaviorOptions(raw = null, {
     shortRest: enabled && shortRest === true,
     batchRest: enabled && batchRest === true,
     actionCooldown: enabled && actionCooldown === true,
+    restLevel,
     restEnabled: enabled && (shortRest === true || batchRest === true)
   };
 }
@@ -404,6 +457,8 @@ export function createHumanRestController({
   shortRestEnabled = true,
   batchRestEnabled = true,
   random = Math.random,
+  nowFn = Date.now,
+  restLevel = DEFAULT_HUMAN_REST_LEVEL,
   shortRestProbability = 0.08,
   shortRestMinMs = 3000,
   shortRestMaxMs = 7000,
@@ -413,12 +468,26 @@ export function createHumanRestController({
   batchRestMaxMs = 30000
 } = {}) {
   const nextRandom = normalizeRandom(random);
+  const readNow = typeof nowFn === "function" ? nowFn : Date.now;
+  const normalizedRestLevel = normalizeHumanRestLevel(restLevel);
+  const budgetProfile = (shortRestEnabled !== false || batchRestEnabled !== false)
+    ? HUMAN_REST_LEVEL_PROFILES[normalizedRestLevel] || null
+    : null;
+  const nextBudgetRestInterval = () => budgetProfile
+    ? randomIntegerBetween(nextRandom, budgetProfile.intervalMin, budgetProfile.intervalMax)
+    : 0;
   const state = {
     enabled: enabled === true,
+    rest_level: normalizedRestLevel,
     short_rest_enabled: enabled === true && shortRestEnabled !== false,
     batch_rest_enabled: enabled === true && batchRestEnabled !== false,
     rest_counter: 0,
     rest_threshold: Math.max(1, Math.floor(Number(batchThresholdBase) || 25) + Math.floor(nextRandom() * Math.max(1, Number(batchThresholdJitter) || 1))),
+    processed_count: 0,
+    candidates_since_last_rest: 0,
+    candidates_until_next_rest: nextBudgetRestInterval(),
+    active_elapsed_ms: 0,
+    last_active_at_ms: Number(readNow()) || 0,
     rest_count: 0,
     total_rest_ms: 0
   };
@@ -427,6 +496,67 @@ export function createHumanRestController({
     state.rest_threshold = Math.max(1, Math.floor(Number(batchThresholdBase) || 25) + Math.floor(nextRandom() * Math.max(1, Number(batchThresholdJitter) || 1)));
   }
 
+  function updateActiveElapsed() {
+    const now = Number(readNow()) || 0;
+    if (state.last_active_at_ms >= 0 && now >= state.last_active_at_ms) {
+      state.active_elapsed_ms += now - state.last_active_at_ms;
+    }
+    state.last_active_at_ms = now;
+    return now;
+  }
+
+  function getBudgetTargetMs() {
+    if (!budgetProfile) return 0;
+    const candidateTarget = state.processed_count * (budgetProfile.targetRestMs / budgetProfile.targetCandidateCount);
+    const elapsedTarget = state.active_elapsed_ms * (budgetProfile.targetRestMs / budgetProfile.targetWindowMs);
+    return Math.max(candidateTarget, elapsedTarget);
+  }
+
+  function chooseBudgetRestPause(debtMs) {
+    const longRest = nextRandom() < budgetProfile.longRestProbability;
+    const minMs = longRest ? budgetProfile.longRestMinMs : budgetProfile.shortRestMinMs;
+    const maxMs = longRest ? budgetProfile.longRestMaxMs : budgetProfile.shortRestMaxMs;
+    const scaleMin = longRest ? 0.75 : 0.38;
+    const scaleMax = longRest ? 1.1 : 0.78;
+    const desiredMs = debtMs * randomBetween(nextRandom, scaleMin, scaleMax);
+    const randomizedMs = randomBetween(nextRandom, minMs, maxMs);
+    const blendedMs = Math.max(minMs, Math.min(maxMs, (desiredMs + randomizedMs) / 2));
+    const maxAllowedMs = Math.max(minMs, debtMs + budgetProfile.maxOverspendMs);
+    return {
+      pauseMs: Math.round(Math.min(blendedMs, maxAllowedMs)),
+      restSize: longRest ? "long" : "short"
+    };
+  }
+
+  async function takeBudgetBreakIfNeeded(sleeper) {
+    state.processed_count += 1;
+    state.candidates_since_last_rest += 1;
+    state.candidates_until_next_rest -= 1;
+    const debtMs = getBudgetTargetMs() - state.total_rest_ms;
+    const intervalDue = state.candidates_until_next_rest <= 0;
+    const forceDue = debtMs >= budgetProfile.forceDebtMs;
+    if (!intervalDue && !forceDue) {
+      return null;
+    }
+    if (debtMs < budgetProfile.minDebtToRestMs) {
+      if (intervalDue) state.candidates_until_next_rest = nextBudgetRestInterval();
+      return null;
+    }
+    const { pauseMs, restSize } = chooseBudgetRestPause(debtMs);
+    await sleeper(pauseMs);
+    const event = {
+      kind: "random_rest",
+      rest_level: normalizedRestLevel,
+      rest_size: restSize,
+      pause_ms: pauseMs,
+      processed_since_last_rest: state.candidates_since_last_rest,
+      rest_budget_debt_ms: Math.round(Math.max(0, debtMs))
+    };
+    state.candidates_since_last_rest = 0;
+    state.candidates_until_next_rest = nextBudgetRestInterval();
+    return event;
+  }
+
   async function takeBreakIfNeeded({ sleepFn = sleep } = {}) {
     if (!state.enabled) {
       return {
@@ -438,18 +568,44 @@ export function createHumanRestController({
       };
     }
     const sleeper = typeof sleepFn === "function" ? sleepFn : sleep;
+    updateActiveElapsed();
+    if (budgetProfile) {
+      const budgetEvent = await takeBudgetBreakIfNeeded(sleeper);
+      const pauseMs = budgetEvent?.pause_ms || 0;
+      if (pauseMs > 0) {
+        state.rest_count += 1;
+        state.total_rest_ms += pauseMs;
+        state.last_active_at_ms = Number(readNow()) || state.last_active_at_ms;
+      }
+      return {
+        enabled: true,
+        rested: Boolean(budgetEvent),
+        pause_ms: pauseMs,
+        rest_level: normalizedRestLevel,
+        rest_counter: state.rest_counter,
+        rest_threshold: state.rest_threshold,
+        processed_count: state.processed_count,
+        candidates_until_next_rest: state.candidates_until_next_rest,
+        active_elapsed_ms: state.active_elapsed_ms,
+        rest_count: state.rest_count,
+        total_rest_ms: state.total_rest_ms,
+        events: budgetEvent ? [budgetEvent] : []
+      };
+    }
     state.rest_counter += 1;
+    state.processed_count += 1;
     const events = [];
     if (state.short_rest_enabled && nextRandom() < Math.max(0, Number(shortRestProbability) || 0)) {
       const pauseMs = Math.round(randomBetween(nextRandom, shortRestMinMs, shortRestMaxMs));
       await sleeper(pauseMs);
-      events.push({ kind: "random_rest", pause_ms: pauseMs });
+      events.push({ kind: "random_rest", rest_level: normalizedRestLevel, pause_ms: pauseMs });
     }
     if (state.batch_rest_enabled && state.rest_counter >= state.rest_threshold) {
       const pauseMs = Math.round(randomBetween(nextRandom, batchRestMinMs, batchRestMaxMs));
       await sleeper(pauseMs);
       events.push({
         kind: "batch_rest",
+        rest_level: normalizedRestLevel,
         pause_ms: pauseMs,
         processed_since_last_batch_rest: state.rest_counter
       });
@@ -460,13 +616,17 @@ export function createHumanRestController({
     if (pauseMs > 0) {
       state.rest_count += events.length;
       state.total_rest_ms += pauseMs;
+      state.last_active_at_ms = Number(readNow()) || state.last_active_at_ms;
     }
     return {
       enabled: true,
       rested: events.length > 0,
       pause_ms: pauseMs,
+      rest_level: normalizedRestLevel,
       rest_counter: state.rest_counter,
       rest_threshold: state.rest_threshold,
+      processed_count: state.processed_count,
+      active_elapsed_ms: state.active_elapsed_ms,
       rest_count: state.rest_count,
       total_rest_ms: state.total_rest_ms,
       events

package/src/domains/chat/run-service.js

@@ -743,11 +743,12 @@ export async function runChatWorkflow({
     safeClickPointEnabled: effectiveHumanBehavior.clickMovement,
     actionCooldownEnabled: effectiveHumanBehavior.actionCooldown
   });
-  const humanRestController = createHumanRestController({
-    enabled: effectiveHumanRestEnabled,
-    shortRestEnabled: effectiveHumanBehavior.shortRest,
-    batchRestEnabled: effectiveHumanBehavior.batchRest
-  });
+  const humanRestController = createHumanRestController({
+    enabled: effectiveHumanRestEnabled,
+    shortRestEnabled: effectiveHumanBehavior.shortRest,
+    batchRestEnabled: effectiveHumanBehavior.batchRest,
+    restLevel: effectiveHumanBehavior.restLevel
+  });
   const normalizedDetailSource = normalizeDetailSource(detailSource);
   const normalizedScreeningMode = normalizeScreeningMode(screeningMode);
   const useLlmScreening = normalizedScreeningMode !== "deterministic";
@@ -966,11 +967,12 @@ export async function runChatWorkflow({
       context_recoveries: contextRecoveryAttempts,
       list_end_reason: listEndReason,
       viewport_checks: viewportGuard.getStats().checks,
-      viewport_recoveries: viewportGuard.getStats().recoveries,
-      human_behavior_enabled: effectiveHumanBehavior.enabled,
-      human_behavior_profile: effectiveHumanBehavior.profile,
-      human_rest_enabled: effectiveHumanRestEnabled,
-      human_rest_count: humanRestController.getState().rest_count,
+      viewport_recoveries: viewportGuard.getStats().recoveries,
+      human_behavior_enabled: effectiveHumanBehavior.enabled,
+      human_behavior_profile: effectiveHumanBehavior.profile,
+      human_rest_level: effectiveHumanBehavior.restLevel,
+      human_rest_enabled: effectiveHumanRestEnabled,
+      human_rest_count: humanRestController.getState().rest_count,
       human_rest_ms: humanRestController.getState().total_rest_ms,
       last_human_event: lastHumanEvent
     });
@@ -1771,11 +1773,12 @@ export async function runChatWorkflow({
       context_recoveries: contextRecoveryAttempts,
       list_end_reason: listEndReason || null,
       viewport_checks: viewportGuard.getStats().checks,
-      viewport_recoveries: viewportGuard.getStats().recoveries,
-      human_behavior_enabled: effectiveHumanBehavior.enabled,
-      human_behavior_profile: effectiveHumanBehavior.profile,
-      human_rest_enabled: effectiveHumanRestEnabled,
-      human_rest_count: humanRestController.getState().rest_count,
+      viewport_recoveries: viewportGuard.getStats().recoveries,
+      human_behavior_enabled: effectiveHumanBehavior.enabled,
+      human_behavior_profile: effectiveHumanBehavior.profile,
+      human_rest_level: effectiveHumanBehavior.restLevel,
+      human_rest_enabled: effectiveHumanRestEnabled,
+      human_rest_count: humanRestController.getState().rest_count,
       human_rest_ms: humanRestController.getState().total_rest_ms,
       last_human_event: lastHumanEvent,
       last_candidate_id: screeningCandidate.id || null,
@@ -1815,10 +1818,11 @@ export async function runChatWorkflow({
         compactResult.human_rest = restResult;
         addTiming(compactResult.timings, "human_rest_ms", restElapsed);
         compactResult.timings.total_ms = Date.now() - candidateStarted;
-        runControl.updateProgress({
-          human_rest_enabled: effectiveHumanRestEnabled,
-          human_rest_count: humanRestController.getState().rest_count,
-          human_rest_ms: humanRestController.getState().total_rest_ms,
+        runControl.updateProgress({
+          human_rest_enabled: effectiveHumanRestEnabled,
+          human_rest_level: effectiveHumanBehavior.restLevel,
+          human_rest_count: humanRestController.getState().rest_count,
+          human_rest_ms: humanRestController.getState().total_rest_ms,
           human_rest_last: restResult,
           context_recoveries: contextRecoveryAttempts,
           last_human_event: lastHumanEvent
@@ -1958,12 +1962,13 @@ export function createChatRunService({
         list_settle_ms: listSettleMs,
         list_fallback_point: listFallbackPoint,
         online_resume_button_timeout_ms: onlineResumeButtonTimeoutMs,
-        image_output_dir: imageOutputDir || "",
-        human_behavior_enabled: effectiveHumanBehavior.enabled,
-        human_behavior_profile: effectiveHumanBehavior.profile,
-        human_behavior: effectiveHumanBehavior,
-        human_rest_enabled: effectiveHumanRestEnabled
-      },
+        image_output_dir: imageOutputDir || "",
+        human_behavior_enabled: effectiveHumanBehavior.enabled,
+        human_behavior_profile: effectiveHumanBehavior.profile,
+        human_behavior: effectiveHumanBehavior,
+        human_rest_level: effectiveHumanBehavior.restLevel,
+        human_rest_enabled: effectiveHumanRestEnabled
+      },
       progress: {
         card_count: 0,
         target_count: targetPassCount || (processUntilListEnd ? "all" : processedLimit),
@@ -1978,10 +1983,11 @@ export function createChatRunService({
         requested: 0,
         request_satisfied: 0,
         request_skipped: 0,
-        context_recoveries: 0,
-        human_behavior_enabled: effectiveHumanBehavior.enabled,
-        human_behavior_profile: effectiveHumanBehavior.profile,
-        human_rest_enabled: effectiveHumanRestEnabled,
+        context_recoveries: 0,
+        human_behavior_enabled: effectiveHumanBehavior.enabled,
+        human_behavior_profile: effectiveHumanBehavior.profile,
+        human_rest_level: effectiveHumanBehavior.restLevel,
+        human_rest_enabled: effectiveHumanRestEnabled,
         human_rest_count: 0,
         human_rest_ms: 0,
         last_human_event: null

package/src/domains/recommend/run-service.js

@@ -690,7 +690,8 @@ export async function runRecommendWorkflow({
   const humanRestController = createHumanRestController({
     enabled: effectiveHumanRestEnabled,
     shortRestEnabled: effectiveHumanBehavior.shortRest,
-    batchRestEnabled: effectiveHumanBehavior.batchRest
+    batchRestEnabled: effectiveHumanBehavior.batchRest,
+    restLevel: effectiveHumanBehavior.restLevel
   });
   const normalizedFilter = normalizeFilter(filter);
   const normalizedPostAction = normalizeRecommendPostAction(postAction) || "none";
@@ -790,6 +791,7 @@ export async function runRecommendWorkflow({
       viewport_recoveries: viewportGuard.getStats().recoveries,
       human_behavior_enabled: effectiveHumanBehavior.enabled,
       human_behavior_profile: effectiveHumanBehavior.profile,
+      human_rest_level: effectiveHumanBehavior.restLevel,
       human_rest_enabled: effectiveHumanRestEnabled,
       human_rest_count: humanRestState.rest_count,
       human_rest_ms: humanRestState.total_rest_ms,
@@ -1509,6 +1511,7 @@ export async function runRecommendWorkflow({
         addTiming(compactResult.timings, "human_rest_ms", restElapsed);
         compactResult.timings.total_ms = Date.now() - candidateStarted;
         updateRecommendProgress({
+          human_rest_level: effectiveHumanBehavior.restLevel,
           human_rest_last: restResult
         });
       }
@@ -1651,6 +1654,7 @@ export function createRecommendRunService({
         human_behavior_enabled: effectiveHumanBehavior.enabled,
         human_behavior_profile: effectiveHumanBehavior.profile,
         human_behavior: effectiveHumanBehavior,
+        human_rest_level: effectiveHumanBehavior.restLevel,
         human_rest_enabled: effectiveHumanRestEnabled
       },
       progress: {
@@ -1670,6 +1674,7 @@ export function createRecommendRunService({
         context_recoveries: 0,
         human_behavior_enabled: effectiveHumanBehavior.enabled,
         human_behavior_profile: effectiveHumanBehavior.profile,
+        human_rest_level: effectiveHumanBehavior.restLevel,
         human_rest_enabled: effectiveHumanRestEnabled,
         human_rest_count: 0,
         human_rest_ms: 0,

package/src/domains/recruit/run-service.js

@@ -395,7 +395,8 @@ export async function runRecruitWorkflow({
   const humanRestController = createHumanRestController({
     enabled: effectiveHumanRestEnabled,
     shortRestEnabled: effectiveHumanBehavior.shortRest,
-    batchRestEnabled: effectiveHumanBehavior.batchRest
+    batchRestEnabled: effectiveHumanBehavior.batchRest,
+    restLevel: effectiveHumanBehavior.restLevel
   });
   const normalizedSearchParams = normalizeSearchParams(searchParams);
   const normalizedScreeningMode = normalizeScreeningMode(screeningMode);
@@ -485,6 +486,7 @@ export async function runRecruitWorkflow({
       viewport_recoveries: viewportGuard.getStats().recoveries,
       human_behavior_enabled: effectiveHumanBehavior.enabled,
       human_behavior_profile: effectiveHumanBehavior.profile,
+      human_rest_level: effectiveHumanBehavior.restLevel,
       human_rest_enabled: effectiveHumanRestEnabled,
       human_rest_count: humanRestState.rest_count,
       human_rest_ms: humanRestState.total_rest_ms,
@@ -1101,6 +1103,7 @@ export async function runRecruitWorkflow({
         addTiming(compactResult.timings, "human_rest_ms", restElapsed);
         compactResult.timings.total_ms = Date.now() - candidateStarted;
         updateRecruitProgress({
+          human_rest_level: effectiveHumanBehavior.restLevel,
           human_rest_last: restResult
         });
       }
@@ -1223,6 +1226,7 @@ export function createRecruitRunService({
         human_behavior_enabled: effectiveHumanBehavior.enabled,
         human_behavior_profile: effectiveHumanBehavior.profile,
         human_behavior: effectiveHumanBehavior,
+        human_rest_level: effectiveHumanBehavior.restLevel,
         human_rest_enabled: effectiveHumanRestEnabled
       },
       progress: {
@@ -1239,6 +1243,7 @@ export function createRecruitRunService({
         context_recoveries: 0,
         human_behavior_enabled: effectiveHumanBehavior.enabled,
         human_behavior_profile: effectiveHumanBehavior.profile,
+        human_rest_level: effectiveHumanBehavior.restLevel,
         human_rest_enabled: effectiveHumanRestEnabled,
         human_rest_count: 0,
         human_rest_ms: 0,

package/src/index.js

@@ -87,8 +87,9 @@ import {
 } from "./run-state.js";
 
 const require = createRequire(import.meta.url);
-const { version: SERVER_VERSION } = require("../package.json");
-
+const { version: SERVER_VERSION } = require("../package.json");
+
+const TOOL_PREPARE_RUN = "prepare_recommend_pipeline_run";
 const TOOL_START_RUN = "start_recommend_pipeline_run";
 const TOOL_GET_RUN = "get_recommend_pipeline_run";
 const TOOL_CANCEL_RUN = "cancel_recommend_pipeline_run";
@@ -308,7 +309,17 @@ function createHumanBehaviorInputSchema(description = "可选，启用可靠性
       listScrollJitter: { type: "boolean" },
       shortRest: { type: "boolean" },
       batchRest: { type: "boolean" },
-      actionCooldown: { type: "boolean" }
+      actionCooldown: { type: "boolean" },
+      restLevel: {
+        type: "string",
+        enum: ["low", "medium", "high"],
+        description: "本次 run 的休息强度：low 保持旧策略；medium 约 5 小时/700 人累计休息 30 分钟；high 约 5 小时/700 人累计休息 1 小时"
+      },
+      rest_level: {
+        type: "string",
+        enum: ["low", "medium", "high"],
+        description: "兼容字段；优先使用 restLevel"
+      }
     },
     additionalProperties: false,
     description
@@ -1079,10 +1090,15 @@ function createToolsSchema() {
       description: "CDP-only 读取 Boss 推荐页岗位下拉框，返回所有可用岗位完整名称，方便 cron/一次性任务提前填写 job 参数。不会启动筛选任务。",
       inputSchema: createListRecommendJobsInputSchema()
     },
+    {
+      name: TOOL_PREPARE_RUN,
+      description: "只校验 Boss 推荐页流水线参数是否已可用于 cron/一次性任务；不会启动筛选任务。只有返回 READY/cron_ready=true 后才应创建定时任务。",
+      inputSchema: createRunInputSchema()
+    },
     {
       name: TOOL_START_RUN,
       description: "异步启动 Boss 推荐页流水线（含同步门禁预检）；只有在前置确认与页面就绪通过后才返回 run_id。",
-      inputSchema: createRunInputSchema()
+      inputSchema: createRunInputSchema()
     },
     {
       name: TOOL_GET_RUN,
@@ -2608,6 +2624,8 @@ async function handleRequest(message, workspaceRoot) {
       let payload;
       if (toolName === TOOL_LIST_RECOMMEND_JOBS) {
         payload = await listRecommendJobsTool({ workspaceRoot, args });
+      } else if (toolName === TOOL_PREPARE_RUN) {
+        payload = prepareRecommendPipelineRunTool({ workspaceRoot, args });
       } else if (toolName === TOOL_START_RUN) {
         payload = await handleStartRunTool({ workspaceRoot, args });
       } else if (toolName === TOOL_GET_RUN) {

package/src/recommend-mcp.js

@@ -1464,15 +1464,21 @@ async function startRecommendPipelineRunInternal(args = {}, { workspaceRoot = ""
   };
 }
 
-export function prepareRecommendPipelineRunTool({ workspaceRoot = "", args = {} } = {}) {
-  const prepared = prepareRecommendPipelineStart(args, { workspaceRoot });
-  if (prepared.response) return prepared.response;
-  const { parsed, normalized } = prepared;
-  return {
-    status: "READY",
-    review: parsed.review,
-    post_action: {
-      requested: normalized.postAction,
+export function prepareRecommendPipelineRunTool({ workspaceRoot = "", args = {} } = {}) {
+  const prepared = prepareRecommendPipelineStart(args, { workspaceRoot });
+  if (prepared.response) {
+    return {
+      ...prepared.response,
+      cron_ready: false
+    };
+  }
+  const { parsed, normalized } = prepared;
+  return {
+    status: "READY",
+    cron_ready: true,
+    review: parsed.review,
+    post_action: {
+      requested: normalized.postAction,
       execute_post_action: args.dry_run_post_action === true ? false : args.execute_post_action !== false,
       max_greet_count: normalized.maxGreetCount
     },

package/src/recruit-mcp.js

@@ -665,7 +665,17 @@ function createHumanBehaviorInputSchema(description = "可选，search/recruit
       listScrollJitter: { type: "boolean" },
       shortRest: { type: "boolean" },
       batchRest: { type: "boolean" },
-      actionCooldown: { type: "boolean" }
+      actionCooldown: { type: "boolean" },
+      restLevel: {
+        type: "string",
+        enum: ["low", "medium", "high"],
+        description: "本次 run 的休息强度：low 保持旧策略；medium 约 5 小时/700 人累计休息 30 分钟；high 约 5 小时/700 人累计休息 1 小时"
+      },
+      rest_level: {
+        type: "string",
+        enum: ["low", "medium", "high"],
+        description: "兼容字段；优先使用 restLevel"
+      }
     },
     additionalProperties: false,
     description