dsh-plugin-t-expert 0.2.6 → 0.2.9

package/data/source.json

@@ -1,6 +1,6 @@
 {
-  "updatedAt": "2026-09-12 15:56:06 +0800",
-  "expertFiles": 314,
+  "updatedAt": "2026-09-12 22:44:36 +0800",
+  "expertFiles": 316,
   "divisions": [
     "academic",
     "company",

package/data/t-team.config.json

@@ -1,5 +1,4 @@
 {
-  "stateDir": ".agent-teams",
   "memberProvider": "spawn",
   "profiles": {
     "t-team-feature": {
@@ -1838,6 +1837,22 @@
           "executionPrompt": "# 🏥 医疗账单与编码专员\n\n> \"医疗账单不是行政开销——它是每一家医疗机构的财务引擎。clean claim 率（一次通过率）哪怕提升 2%，对一家中型机构都可能意味着几十万美元的收入回收。把编码做对。把 claim 做干净。把钱拿到手。\"\n\n## 🧠 你的身份与记忆\n\n你是 **医疗账单与编码专员**——一位持证的收入周期管理（revenue cycle management）专家，在 ICD-10-CM/PCS 诊断编码、CPT 操作编码、HCPCS Level II 编码、claim 提交、denial 管理、payer 合同谈判、合规审计，以及覆盖医师诊所、医院、门诊机构和专科诊所的收入周期优化方面具有深厚造诣。你曾为因 denial 损失 15% 收入的机构重建收入周期，实施过经受住 payer 审计的编码合规项目，谈下过为年收入增加七位数的合同费率。你深知准确编码既是财务要务，也是法律义务——并以此态度对待它。\n\n你记得：\n- 服务提供方的专科、payer 构成（payer mix）和机构类型\n- 当前的 clean claim 率、denial 率和 AR（应收账款）天数\n- 在用的 payer 合同及其费率表（fee schedule）\n- 未结的被拒 claim 及其当前申诉状态\n- 合规审计发现的问题及整改状态\n- 该提供方专科特有的编码政策与文档要求\n\n## 🎯 你的核心使命\n\n通过确保准确编码、干净的 claim 提交、强势的 denial 管理和持续的收入周期改进，最大化收入回收、最小化合规风险——让医疗服务提供方能专注于患者诊疗，而账单引擎始终以巅峰状态运转。\n\n你的工作覆盖完整收入周期：\n- **医疗编码**：ICD-10-CM/PCS、CPT、HCPCS Level II——准确、合规、优化\n- **费用采集（Charge Capture）**：superbill（费用清单）审核、费用录入、费率表管理\n- **Claim 提交**：claim 校验（scrubbing）、电子提交、清算所（clearinghouse）管理\n- **Denial 管理**：denial 分析、申诉、根因整改\n- **应收账款（AR）**：AR 账龄、跟进流程、坏账核销管理\n- **Payer 关系**：合同分析、资质认证（credentialing）支持、事前授权（prior authorization）\n- **合规**：编码审计、文档改进、OIG 指南遵循\n- **报表**：KPI 仪表盘、payer 绩效分析、收入周期基准对标\n\n---\n\n## 🚨 你必须遵守的关键规则\n\n1. **只编码记录了的内容——绝不编码假设的内容。** 编码必须反映医师在病历中记录的内容。绝不臆断诊断、绝不高编（upcode）操作、绝不为未记录的病情赋码。那是欺诈。\n2. **ICD-10 要求特异性。** ICD-10 要求达到可用的最高特异性。\"糖尿病\"是不够的——\"2 型糖尿病伴糖尿病性慢性肾病 3 期\"才是。未特指（unspecified）编码应是最后手段，而非默认选项。\n3. **每一项计费服务都必须有医疗必要性（medical necessity）支撑。** 每张 claim 都必须有医疗必要性支撑——即记录在案的、说明该服务为何必需的临床理由。无记录医疗必要性的服务会被 denial，若被审计，还可能构成 false claims（虚假理赔）。\n4. **绝不为未提供的服务计费。** 为未执行的服务计费——无论本意如何、是否已排程——都是欺诈。计费前先核实服务文档。\n\n（以上为该专家的人格摘录，来自 T专家 名册；完整正文可用 summon_t_expert 获取。）"
         }
       ]
+    },
+    "dsht": {
+      "description": "Dsh专家",
+      "taskPlanning": "captain",
+      "members": [
+        {
+          "name": "DSH 项目专家",
+          "role": "engineering-deepseek-harness-project-expert",
+          "executionPrompt": "# DeepSeek Harness Project Expert Agent\n\nYou are **DeepSeek Harness Project Expert**, permanently assigned to one checkout: DeepSeek Harness, the all-plugin Cordis agent harness. You do not answer about \"an agent framework in general\" — you answer about *this* repository, from *this* repository, the way its maintainers do. Model adapters, the tool registry, the session log, and even the agent loop are plugins; the extension move is always \"mount a plugin beside the core\", never \"patch the core\", because there is no core to patch. Every registration is an effect that unwinds when its plugin unloads.\n\n## 🧠 Your Identity & Memory\n\n- **Role**: Maintainer-level expert for a pnpm-workspace, all-plugin TypeScript harness — architecture, boot model, extension points, per-language rules, and quality gates\n- **Personality**: Plane-first, seam-literate, allergic to \"just put it in the loop\", precise about which surface (source or artifact) a claim belongs to\n- **Memory**: You remember every change that landed on the wrong plane, every registration that wasn't an effect and therefore leaked on reload, every model-visible input that couldn't be rebuilt from the session log, every `interface` mistaken for a Service Definition, every `catch` that swallowed a failure the repository's own rules say must be loud\n\n（以上为该专家的人格摘录，来自 T专家 名册；完整正文可用 summon_t_expert 获取。）"
+        },
+        {
+          "name": "TS/npm 开发维护者",
+          "role": "engineering-typescript-npm-stack-maintainer",
+          "executionPrompt": "# TypeScript / npm Stack Maintainer Agent\n\nYou are **TypeScript / npm Stack Maintainer**, the engineer who owns a repository whose language bar reads TypeScript at the top and then a thin tail of everything else — a few percent of CSS, a sliver of Python, some JavaScript glue, and languages that round to 0.0% until they break the build. You are not a TypeScript-only specialist who treats the other 3% as someone else's problem. You are the one who knows that the 3% is where releases actually fail.\n\n## 🧠 Your Identity & Memory\n\n- **Role**: Full-surface maintainer for a TS-dominant, npm-published codebase — source, build, bundle, tooling, packaging, and release\n- **Personality**: Type-strict, packaging-paranoid, allergic to \"it works on my machine\", respectful of small scripts that quietly hold everything together\n- **Memory**: You remember every release broken by a missing `files` entry, every `exports` map that resolved in dev and 404'd after install, every shell script that was fine until it ran under `sh` instead of `bash`, every native dependency that compiled locally and vanished in CI\n- **Experience**: You've shipped packages where 96.9% TypeScript took 3% of the debugging time. You stopped believing language percentages say anything about where the risk lives.\n\n## 🎯 Your Core Mission\n\n### Ship TypeScript that survives strict mode and review\n- All new logic is TypeScript with `strict` on — no `any` escape hatches, no non-null assertions standing in for real narrowing\n\n（以上为该专家的人格摘录，来自 T专家 名册；完整正文可用 summon_t_expert 获取。）"
+        }
+      ]
     }
   }
 }

package/data/team-profiles.py

@@ -1,25 +1,26 @@
 #!/usr/bin/env python3
-"""把 T专家 的小队定义（teams.json）编译成 AgentTeams 的 profiles 配置。
+"""把 T专家 的小队定义（teams.json）编译成内置团队引擎的 profiles 配置。
 
-职责分工：T专家 管「人」（名册 + 中文 + 人格），AgentTeams 管「队」（成员生命周期、任务 DAG、
+职责分工：T专家 管「人」（名册 + 中文 + 人格），内置引擎 管「队」（成员生命周期、任务 DAG、
 调度、邮箱、面板）。本脚本只做**数据转换**：读 teams.json + experts/ + zh/，把每位专家的
-中文名与人格正文写成 profile 成员的 executionPrompt，然后写进 DSH profile 的
-`cordis.patch.yml`（用标记块包裹，幂等可重放）。
+中文名与人格正文写成 profile 成员的 executionPrompt，然后写进 `t-team.config.json`
+（引擎由 T专家 在 apply 时挂载并读这份配置）。
 
 用法：
-    python3 team-profiles.py                 # 生成并写入 profile patch（自动备份）
-    python3 team-profiles.py --dry-run       # 只打印校验与将要写入的内容
+    python3 team-profiles.py                 # 生成并写入 t-team.config.json（+ teams.resolved.json）
+    python3 team-profiles.py --dry-run       # 只打印校验与**将要写入的内容**（预览 == 产物）
     python3 team-profiles.py --mode full     # 成员人格用完整正文（默认 summary 摘录）
-    python3 team-profiles.py --no-write      # 只输出片段到 stdout，不动文件
+    python3 team-profiles.py --no-write      # 同上，但不动文件
 
 注意（两个已确认的坑）：
-  1) profile patch 行的 config 是**整体替换**而非深合并，所以这里会把 stateDir / memberProvider
-     一起写全，避免抹掉 AgentTeams 自带默认值。
+  1) **不要把 stateDir 写进 t-team.config.json**：状态目录的唯一真源是插件侧 Config.stateDir，
+     数据文件里再留一份就会出现「文件里一个值、实际生效另一个值」的双来源（审计 D-1）。
   2) AgentTeams 的 profile key 只支持 ASCII（中文 key 不会生成命令别名），中文放 description/aliases。
 """
 from __future__ import annotations
 
 import argparse
+import copy
 import json
 import os
 import re
@@ -43,7 +44,10 @@ MARKER_START = "# >>> t-team teams（由 team-profiles.py 生成，勿手改；
 MARKER_END = "# <<< t-team teams"
 
 MAX_PROFILES = 48          # 引擎侧 lib/teams/profiles.js 已把 MAX_TEAM_PROFILES 调到 48（原默认 16）
-MAX_MEMBERS = 8            # 一个 profile 的成员数上限（经验值：队长上下文压力）
+# 一个 profile 的成员数上限（经验值：队长上下文压力）。
+# 只是**默认**：插件侧 Config.maxMembers 经 --max-members 覆盖它，两边同一个真源；
+# 直接跑本脚本（不带该参数）时行为与从前完全一致。
+MAX_MEMBERS = 8
 MAX_SEED_TASKS = 32        # AgentTeams: MAX_PROFILE_TASKS
 PROFILE_KEY_RE = re.compile(r"^[a-z0-9]+(?:-[a-z0-9]+)*$")
 NAME_FIELD = re.compile(r"^name\s*:", re.M)
@@ -107,10 +111,10 @@ def load_roster(experts_root: str, zh_root: str) -> dict[str, dict]:
 
 
 def build_members(slugs: list[str], roster: dict[str, dict], mode: str, budget: int,
-                  errors: list[str], where: str) -> list[dict]:
+                  errors: list[str], where: str, max_members: int = MAX_MEMBERS) -> list[dict]:
     members, used_names = [], {}
-    if len(slugs) > MAX_MEMBERS:
-        errors.append(f"{where}: 成员 {len(slugs)} 位超过上限 {MAX_MEMBERS}")
+    if len(slugs) > max_members:
+        errors.append(f"{where}: 成员 {len(slugs)} 位超过上限 {max_members}")
     for slug in slugs:
         expert = roster.get(slug)
         if expert is None:
@@ -160,48 +164,31 @@ def emit_member(member: dict, indent: str) -> str:
     return "\n".join(lines)
 
 
-def render_block(profiles: dict[str, dict]) -> str:
-    lines = [MARKER_START, "- id: agent-teams", "  config:",
-             "    stateDir: .agent-teams        # 与 AgentTeams 自带默认值保持一致（patch 的 config 是整体替换）",
-             "    memberProvider: spawn",
-             "    profiles:"]
-    for key, profile in profiles.items():
-        lines.append(f"      {key}:")
-        lines.append(f"        description: {yaml_scalar(profile['description'])}")
-        lines.append(f"        taskPlanning: {profile['taskPlanning']}")
-        if profile.get("tasks"):
-            lines.append("        tasks:")
-            for task in profile["tasks"]:
-                lines.append(f"          - id: {yaml_scalar(task['id'])}")
-                lines.append(f"            subject: {yaml_scalar(task['subject'])}")
-                if task.get("description"):
-                    lines.append(f"            description: {yaml_scalar(task['description'])}")
-                if task.get("assignee"):
-                    lines.append(f"            assignee: {yaml_scalar(task['assignee'])}")
-                deps = task.get("dependencies") or []
-                lines.append("            dependencies: [" + ", ".join(yaml_scalar(d) for d in deps) + "]")
-        lines.append("        members:")
-        for member in profile["members"]:
-            lines.append(emit_member(member, "          "))
-    lines.append(MARKER_END)
-    return "\n".join(lines) + "\n"
-
-
-def write_patch(patch_path: str, block: str, dry_run: bool) -> str:
-    """把标记块替换进 profile patch（保留其它条目），返回备份路径。"""
-    old = open(patch_path, encoding="utf-8").read() if os.path.isfile(patch_path) else "[]\n"
-    pattern = re.compile(re.escape(MARKER_START) + r".*?" + re.escape(MARKER_END) + r"\n?", re.S)
-    if pattern.search(old):
-        new = pattern.sub(block, old)
-    else:
-        base = old if old.strip() and old.strip() != "[]" else ""
-        new = (base.rstrip("\n") + "\n\n" if base else "") + block
-    backup = f"{patch_path}.bak.{datetime.now():%Y%m%d-%H%M%S}"
-    if dry_run:
-        return backup
-    shutil.copy2(patch_path, backup) if os.path.isfile(patch_path) else None
-    open(patch_path, "w", encoding="utf-8").write(new)
-    return backup
+def build_engine_config(spec: dict, profiles: dict[str, dict]) -> dict:
+    """构造**真正会写下去**的引擎配置（`t-team.config.json` 的那份对象）。
+
+    预览与落盘共用这一个构造点，两者不可能再漂移 —— 这是「预览必须等于产物」的实现方式。
+
+    注意：**不要**把 stateDir 写进这份 JSON。状态目录的唯一真源是插件侧 Config.stateDir
+    （lib/index.js 用它挂载引擎、找团队状态、跑 t_team_plan_check）。这里再写一份，就会出现
+    「数据文件里一个值、实际生效另一个值」的双来源：改开 Config.stateDir 的部署两边失明且无报错
+    （审计 D-1）。历史数据文件里的该字段已删除，且插件不再读它。
+    """
+    return {
+        "memberProvider": (spec.get("engine") or {}).get("memberProvider", "spawn"),
+        "profiles": {
+            key: {
+                "description": profile["description"],
+                "taskPlanning": profile["taskPlanning"],
+                "members": [
+                    {"name": m["name"], "role": m["role"], "executionPrompt": m["executionPrompt"]}
+                    for m in profile["members"]
+                ],
+                **({"tasks": profile["tasks"]} if profile.get("tasks") else {}),
+            }
+            for key, profile in profiles.items()
+        },
+    }
 
 
 def remove_patch_block(patch_path: str, dry_run: bool) -> bool:
@@ -220,6 +207,16 @@ def remove_patch_block(patch_path: str, dry_run: bool) -> bool:
     return True
 
 
+def read_engine_state_dir(engine_config_path: str) -> object:
+    """读盘上那份产物里的 stateDir 死字段（只为让预览如实说明「它会被丢弃」；读不到就返回 None）。"""
+    if not os.path.isfile(engine_config_path):
+        return None
+    try:
+        return json.load(open(engine_config_path, encoding="utf-8")).get("stateDir")
+    except (OSError, json.JSONDecodeError):
+        return None
+
+
 def main() -> int:
     ap = argparse.ArgumentParser()
     ap.add_argument("--teams", default=DEFAULT_TEAMS)
@@ -227,10 +224,15 @@ def main() -> int:
     ap.add_argument("--zh", default=DEFAULT_ZH)
     ap.add_argument("--patch", default=DEFAULT_PATCH)
     ap.add_argument("--mode", choices=["summary", "full", "reference"], help="覆盖默认人格模式")
+    ap.add_argument("--max-members", type=int, default=MAX_MEMBERS,
+                    help=f"一个 profile 的成员数上限（默认 {MAX_MEMBERS}；插件侧由 Config.maxMembers 透传）")
     ap.add_argument("--dry-run", action="store_true")
     ap.add_argument("--no-write", action="store_true")
     args = ap.parse_args()
 
+    if args.max_members < 1:
+        print(f"错误：--max-members 必须是正整数，收到 {args.max_members}", file=sys.stderr)
+        return 1
     if not os.path.isfile(args.teams):
         print(f"错误：找不到小队定义 {args.teams}", file=sys.stderr)
         return 1
@@ -267,7 +269,7 @@ def main() -> int:
                 aliases[alias] = key
         mode = args.mode or profile.get("personaMode") or defaults.get("personaMode") or "summary"
         budget = int(profile.get("personaBudget") or defaults.get("personaBudget") or 1600)
-        members = build_members(list(profile.get("members") or []), roster, mode, budget, errors, key)
+        members = build_members(list(profile.get("members") or []), roster, mode, budget, errors, key, args.max_members)
         if not members:
             errors.append(f"{key}: 没有任何有效成员")
             continue
@@ -303,32 +305,32 @@ def main() -> int:
             print("  ✗", error, file=sys.stderr)
         return 1
 
-    block = render_block({k: v for k, v in profiles.items()})
+    # 引擎配置：T专家 在 apply 时读它并 ctx.plugin(内置引擎, 该配置)
+    engine_config_path = os.path.join(os.path.dirname(os.path.abspath(args.teams)), "t-team.config.json")
+    engine_config = build_engine_config(spec, profiles)
+
     if args.dry_run or args.no_write:
-        print("\n" + "=" * 60 + "\n" + block)
+        # 预览必须等于产物：这里打印的就是上面那个 engine_config 对象。
+        # 旧实现在这里打印一段带 `- id: agent-teams` / `stateDir: .agent-teams` 的**遗留 patch 块**，
+        # 既与实际产物不符，又与 build_engine_config 的「不要把 stateDir 写进这份 JSON」注释自相矛盾
+        # —— 预览是给人看的，看到的东西必须等于真正会写下去的东西（复核 F-1）。
+        preview = copy.deepcopy(engine_config)
+        legacy_state_dir = read_engine_state_dir(engine_config_path)
+        # 预览的键集合必须与产物一致，所以 stateDir **不作为产物键**出现；
+        # 盘上若还留着历史死字段，另起一行如实说明它会被丢弃（不塞进 JSON 里假装是产物）。
+        print("\n" + "=" * 60)
+        print(f"预览：将写入 {engine_config_path}（内容与该文件实际写入的完全一致）：\n")
+        print(json.dumps(preview, ensure_ascii=False, indent=2))
+        print("\nstateDir：不是本文件的产物字段（预览与产物都不含它）。"
+              "状态目录的唯一真源是插件侧 Config.stateDir（审计 D-1）；"
+              f"历史遗留字段{'会被丢弃（盘上当前值 ' + repr(legacy_state_dir) + '）' if legacy_state_dir is not None else '已不在盘上'}。")
         if args.dry_run:
             print("（dry-run，未写盘）")
         return 0
 
-    # 引擎配置：T专家 在 apply 时读它并 ctx.plugin(内置引擎, 该配置)
-    engine_config_path = os.path.join(os.path.dirname(os.path.abspath(args.teams)), "t-team.config.json")
-    json.dump({
-        "stateDir": (spec.get("engine") or {}).get("stateDir", ".agent-teams"),
-        "memberProvider": (spec.get("engine") or {}).get("memberProvider", "spawn"),
-        "profiles": {
-            key: {
-                "description": profile["description"],
-                "taskPlanning": profile["taskPlanning"],
-                "members": [
-                    {"name": m["name"], "role": m["role"], "executionPrompt": m["executionPrompt"]}
-                    for m in profile["members"]
-                ],
-                **({"tasks": profile["tasks"]} if profile.get("tasks") else {}),
-            }
-            for key, profile in profiles.items()
-        },
-    }, open(engine_config_path, "w", encoding="utf-8"), ensure_ascii=False, indent=2)
-    open(engine_config_path, "a", encoding="utf-8").write("\n")
+    with open(engine_config_path, "w", encoding="utf-8") as handle:
+        json.dump(engine_config, handle, ensure_ascii=False, indent=2)
+        handle.write("\n")
     print(f"  引擎配置    : {engine_config_path}")
 
     removed = remove_patch_block(args.patch, args.dry_run)
@@ -355,7 +357,7 @@ def main() -> int:
     if removed:
         print(f"\n已从 {args.patch} 移除旧的 agent-teams 配置块（团队引擎现由 T专家 内置挂载）")
     print(f"  成员人格体积: {total} 字符（{len(profiles)} 支小队）")
-    print("\n下一步：重启 DSH Desktop 让 profile 生效，然后就能用：")
+    print("\n下一步：重启 DSH Desktop 让新配置生效，然后就能用：")
     for key, profile in profiles.items():
         print(f"  /t {profile['aliases'][0] if profile['aliases'] else key} <目标>     或  /t --profile {key} <目标>")
     return 0