ci3: lint/mypy 第二阶段清理 + 覆盖率 76% -> 78%（门槛不动）

基底：#251 头 174b631（堆叠于 #242 feat-ci-expand 之上），
基线 origin/main aebc6238；本轨不改 #251 已定语义、不动
workflow 门槛、不删检查、不加 skip。

lint（ruff E,F@120）：109 -> 10
- F401 43 -> 0：删未用导入；try/except 双绑定的名字出现
  “打地鼠”（删 except 分支后 try 分支又被标出），两分支
  均已核对正文无引用后清掉；MCP 三模块的 Error 名不删，
  改走 importlib 属性绑定（见下），两路径绑定都保留。
- E702 40 -> 0：全在 host.py main() 的 argparse 段，纯分行，
  tokenize 按分号 token 拆（含 2377 行多行语句只断首个分号）；
  一次 5 空格缩进失误已回滚重做并以 ast + 全量 host 测试验证。
- E501 13 -> 0：host.py 5 项随 E702 分行连带清掉；mcp_server
  3 项把描述拆成相邻字面量（AST 比对 24 条工具描述与原值
  完全一致，注意 ruff 按东亚宽度计长，CJK 算 2）；test_host
  5 项断开 git commit 参数表。
- E402 8 -> 10：生产侧 2 项清掉（db.py _re、github_app
  _base64 移到顶部，之前无引用、无重名）；剩余 10 项全是
  tests/ 的 sys.path 前置 import（6 项旧 + 4 个新测试文件
  沿用同一既定模式），移上去会破坏单文件直跑，有意保留。
- F841 4 -> 0：只去绑定、不删调用（run_task、DB 入队、
  tarfile 上下文、git 提交），各有注释与原测试/新测试覆盖；
  host dispatch_from_branch 新增真实 git 仓库回归测试。
- F811 1 -> 0：server 本地 _row_get 与 peer_messages.row_get
  逐行相同，删本地定义，全部调用点解析到同一 import；
  新增 test_server_row_get（同一对象 + 三路行为）。

mypy（生产范围）：24 -> 0，零新增 type: ignore
- no-redef 9 -> 0：MCP 三模块 try/except 两次 from-import
  改 importlib.import_module + 属性绑定（cloud 优先、flat
  copy 回落、内层 ImportError 同样回落，语义与原来一致；
  flat-copy 子进程测试证实回落后 __module__ 为顶层名）；
  server _row_get 随 F811 消除。
- arg-type 10 -> 0：harness2_tasks shared 改 TypedDict
  逐键精确类型；CodexProbeAdapter 补 no-op validate_tree
  （_adapter_validate 本就跳过缺失方法，行为一致）。
- var-annotated 3 -> 0：sub2api _SHARED_ADMIN/_login_slots、
  verify-multiuser RESULTS 加精确注解。
- assignment/misc 2 -> 0：marker_tools ZoneInfo 先声明
  Any 再 try import（草稿先验 mypy 通过）。

bandit：总数 286 不变，High 仍 4 项，全在 host.py，本轨不改
（属第 3 项白名单轨范围），只做 triage：
- 1319 run_verify：无插值，整串验测命令，管道/&& 须 shell
  语义（新增 test_host_run_verify 断言管道与 && 真实生效）。
- 1497 notify：模板来自算子配置，三处插值全部 shlex.quote，
  证据为既有 test_notify_failure_quotes_shell_metacharacters
  （; touch 与 $(touch ) 不执行）。
- 2099/2103 goal verify：无插值，任务配置验测串，同 run_verify。
- 无新增 High；pip-audit 0 漏洞。

覆盖率（同口径三套件 coverage 计量）：75.96% -> 77.87%
（22392 行/缺 5385 -> 22437 行/缺 4963，显示 76% -> 78%）
- client.py 0% -> 90%（pipe 假子进程 + main 分支）
- run_parallel.py 0% -> 97%、migrate 包装 0% -> 95%
- playwright_mcp 37% -> 62%、imagegen_mcp 47% -> 79%、
  websearch_mcp 51% -> 86%、deploy.py 56% -> 77%
- verify-multiuser 0% -> 27%（仅 check/req helper；main()
  是主机侧真实多用户实弹验证，按门禁不建生产任务、
  不以 mock 冒充，仍为缺口）
- host.py 55% -> 57%（文件 2020 -> 2060 行系 E702 分行
  的计量口径变化，非代码膨胀）
- 未到 85%：门槛保持 75% 不动，不删低覆盖模块出统计口径；
  缺口主力仍是 server.py / worker.py / host.py 大文件，
  下一阶段须针对性深测，非本轨快赢可及。

验证：三套件 1576 -> 1629 项全过（tests 127、cloud/tests
1449 skipped=4、deploy/tests 53）；ruff E9 通过；mypy 0；
bandit High 4（同上）；pip-audit 0 漏洞；试跑产物已删，
提交后工作树干净。

cloud/auth.py

@@ -23,7 +23,6 @@ approve_merge -> maintainer，manage_project -> owner（仅 owner）。
 
 import hashlib
 import secrets
-import sqlite3
 from cloud.db import connect as _db_connect
 import time
 import uuid

cloud/cost.py

@@ -50,7 +50,6 @@ events only.
 
 import json
 import os
-import sqlite3
 from cloud.db import connect as _db_connect
 import time
 from datetime import datetime, timezone

cloud/db.py

@@ -59,6 +59,7 @@ memory.py 的 summaries 完全一致，因为本模块根本不建表）。
 """
 
 import os
+import re as _re
 import sqlite3
 import time
 from abc import ABC, abstractmethod
@@ -436,8 +437,6 @@ _OR_REPLACE_PK = {
 _IDENTITY_TABLES = frozenset(
     {"summaries", "kb_docs", "cost_records", "audit_log"})
 
-import re as _re
-
 _OR_REPLACE_RE = _re.compile(
     r"^\s*INSERT\s+OR\s+REPLACE\s+INTO\s+(\w+)\s*\(([^)]+)\)\s*"
     r"VALUES\s*\((.*)\)\s*$",

cloud/github_app.py

@@ -30,6 +30,7 @@ env secret，App webhook 用库里 secret，验签时两个都试（过渡
 注入 fake http 客户端做单测。
 """
 
+import base64 as _base64
 import hashlib
 import json
 import secrets
@@ -607,8 +608,6 @@ def set_fix_review_task_id(conn, repo, pr_number, comment_id, task_id):
 #（"BEGIN RSA PRIVATE KEY"），openssl 3 默认生成的是 PKCS#8
 #（"BEGIN PRIVATE KEY"），两者都支持。
 
-import base64 as _base64
-
 _SHA256_DIGESTINFO_PREFIX = bytes.fromhex(
     "3031300d060960864801650304020105000420")
 

cloud/harness2_tasks.py

@@ -73,7 +73,7 @@ import urllib.request
 import uuid
 from dataclasses import dataclass, field
 from pathlib import Path
-from typing import Protocol
+from typing import Protocol, TypedDict
 
 from .harness2 import (
     EditResult,
@@ -598,6 +598,12 @@ class CodexProbeAdapter:
     def base_harness(self) -> Tree:
         return dict(self.base)
 
+    def validate_tree(self, harness: Tree) -> None:
+        """探针无项目特有不变式：no-op。harness2._adapter_validate
+        本就对缺失该方法的 adapter 跳过，故 no-op 与跳过行为一致，
+        补上只是让类满足 TaskAdapter 协议（mypy arg-type）。"""
+        return None
+
     def execute(self, label: str, harness: Tree) -> Rollout:
         files = dict(self.source_files)
         files.update(_tree_to_files(harness))
@@ -820,6 +826,20 @@ class CodexReflector:
         return parse_experience(state.result_text or "")
 
 
+class _SharedRoleArgs(TypedDict):
+    """editor/selector/reflector 三角色共用的构造参数：键与三者
+    dataclass 字段一一对应、逐键类型精确。dict 字面量会先塌成
+    dict[str, int | str | None] 再按位置报错（mypy arg-type），
+    TypedDict 展开时才能按字段逐个校验，运行时仍是普通 dict。"""
+
+    environment_id: str | None
+    executor: str | None
+    effort: str | None
+    approval: str | None
+    no_time_limit: bool
+    timeout_seconds: int
+
+
 def run_improvement_via_tasks(
     client: TaskClient,
     *,
@@ -872,9 +892,11 @@ def run_improvement_via_tasks(
         approval=approval,
         no_time_limit=no_time_limit, timeout_seconds=timeout_seconds,
         domain=domain)
-    shared = dict(environment_id=environment_id, executor=executor,
+    shared: _SharedRoleArgs = {
-                  effort=effort, approval=approval, no_time_limit=no_time_limit,
+        "environment_id": environment_id, "executor": executor,
-                  timeout_seconds=timeout_seconds)
+        "effort": effort, "approval": approval,
+        "no_time_limit": no_time_limit, "timeout_seconds": timeout_seconds,
+    }
     editor = CodexEditor(client=client, project_id=project_id, run_tag=run_tag,
                          model=_role_model("editor"), **shared)
     selector = CodexSelector(client=client, project_id=project_id, run_tag=run_tag,

cloud/imagegen_mcp.py

@@ -15,15 +15,23 @@ copying; keys are read from the MCP server process environment
 by imagegen.py, never from the tool arguments.
 """
 
+import importlib
 import json
 import os
 import sys
 
+# cloud 包优先、租户 flat copy 回落（与 playwright_mcp 同款理由：
+# 两次 from-import 会让 mypy 报 no-redef，importlib 属性绑定
+# 在两条路径下都是同一组名字；cloud 导入本身的 ImportError
+# 同样回落，与原来 except 语义一致）。
 try:
-    from cloud.imagegen import ImageGenError, edit, generate
+    _mcp_backend = importlib.import_module("cloud.imagegen")
 except ImportError:  # flat copy in a tenant codex-home
     sys.path.insert(0, os.path.dirname(os.path.abspath(__file__)))
-    from imagegen import ImageGenError, edit, generate  # noqa: E402
+    _mcp_backend = importlib.import_module("imagegen")
+ImageGenError = _mcp_backend.ImageGenError
+edit = _mcp_backend.edit
+generate = _mcp_backend.generate
 
 SERVER_NAME = "imagegen"
 SERVER_VERSION = "0.1.0"

cloud/marker_tools.py

@@ -135,7 +135,13 @@ import re
 import time
 import uuid
 from datetime import datetime, timedelta, timezone
+from typing import Any
 
+# ZoneInfo 在极老 Python 上不存在：先声明为 Any 再 try import，
+# 这样 except 分支的 None 赋值不再是“给 type 赋值”（mypy
+# misc + assignment），运行时仍是“有则类、无则 None”，
+# 下游 `if ZoneInfo is not None` 守卫语义不变。
+ZoneInfo: Any = None
 try:
     from zoneinfo import ZoneInfo
 except Exception:  # pragma: no cover - 极老 Python 无 zoneinfo

cloud/mcp_server.py

@@ -83,7 +83,9 @@ TOOLS = [
     _tool("get_task_files", "列/读任务 worktree 里的文件；不传 path 列顶层，传相对路径读文件或列目录。",
           {"task_id": {"type": "string"},
            "path": {"type": "string"}}, ["task_id"]),
-    _tool("get_task_tree", "取任务 worktree 的结构化目录树（只含名、类型、大小，不读文件内容）；不传 path 列 worktree 根，传相对路径列该子目录；max_depth 1..5 缺省 3，max_entries 1..500 缺省 200，截断时 truncated=True。",
+    _tool("get_task_tree", "取任务 worktree 的结构化目录树（只含名、类型、大小，"
+          "不读文件内容）；不传 path 列 worktree 根，"
+          "传相对路径列该子目录；max_depth 1..5 缺省 3，max_entries 1..500 缺省 200，截断时 truncated=True。",
           {"task_id": {"type": "string"},
            "path": {"type": "string"},
            "max_depth": {"type": "integer"},
@@ -98,7 +100,8 @@ TOOLS = [
           {"task_id": {"type": "string"}}, ["task_id"]),
     _tool("goal_resume_task", "断点续跑一个 goal 任务（服务端状态机校验）。",
           {"task_id": {"type": "string"}}, ["task_id"]),
-    _tool("create_followup", "给进行中任务发后续消息；mode=queued 排队进队尾不打断本轮，direct_send 立即送达且本轮结束后优先开下一轮。",
+    _tool("create_followup", "给进行中任务发后续消息；"
+          "mode=queued 排队进队尾不打断本轮，direct_send 立即送达且本轮结束后优先开下一轮。",
           {"task_id": {"type": "string"},
            "body": {"type": "string"},
            "mode": {"type": "string"}}, ["task_id", "body"]),
@@ -112,7 +115,8 @@ TOOLS = [
           {"project_id": {"type": "string"}}, ["project_id"]),
     _tool("get_models", "取模型一览（价目与默认模型，服务端口径）。",
           {}, []),
-    _tool("list_project_memories", "列出某项目的项目级记忆（此前任务沉淀的经验/约定，新任务会自动注入；top_k 可选，默认 20）。",
+    _tool("list_project_memories", "列出某项目的项目级记忆（此前任务沉淀的经验/约定，"
+          "新任务会自动注入；top_k 可选，默认 20）。",
           {"project_id": {"type": "string"},
            "top_k": {"type": "integer"}}, ["project_id"]),
     _tool("create_project_memory", "给某项目记一条项目级记忆（kind 与 content 必填）。",

cloud/migrations.py

@@ -29,7 +29,6 @@
   本模块不读、不拼连接串、不接受命令行密码参数。
 """
 
-import os
 import re
 import sqlite3
 

cloud/oauth.py

@@ -41,7 +41,6 @@
 import json
 import os
 import secrets
-import sqlite3
 from cloud.db import connect as _db_connect
 import time
 import urllib.parse

cloud/playwright_mcp.py

@@ -23,15 +23,23 @@ each new page gets a fresh Playwright context, so cookies are
 isolated between pages.
 """
 
+import importlib
 import json
 import os
 import sys
 
+# cloud 包优先、租户 flat copy 回落：同一段代码在两种部署形态下
+# 都要跑。importlib 动态取属性而不是两次 from-import：两次
+# from-import 会让 mypy 报 no-redef（cloud.* 与顶层模块名不同），
+# 而属性绑定在两条路径下都是同一组名字；cloud 导入本身的
+# ImportError（含内层缺依赖）同样回落，与原来 except 语义一致。
 try:
-    from cloud.playwright_browser import Browser, BrowserError
+    _mcp_backend = importlib.import_module("cloud.playwright_browser")
 except ImportError:  # flat copy in a tenant codex-home
     sys.path.insert(0, os.path.dirname(os.path.abspath(__file__)))
-    from playwright_browser import Browser, BrowserError  # noqa: E402
+    _mcp_backend = importlib.import_module("playwright_browser")
+Browser = _mcp_backend.Browser
+BrowserError = _mcp_backend.BrowserError
 
 SERVER_NAME = "playwright"
 SERVER_VERSION = "0.1.0"

cloud/server.py

@@ -107,7 +107,6 @@ from cloud.wait_ops import build_pr_diagnostics as _wait_build_diag
 from cloud.wait_ops import create_handoff as _handoff_create
 from cloud.wait_ops import decode_wait_cursor as _wait_decode_cursor
 from cloud.wait_ops import encode_wait_cursor as _wait_encode_cursor
-from cloud.wait_ops import ensure_handoffs_table as _handoff_ensure
 from cloud.wait_ops import poll_for_first_change as _wait_poll_first
 from cloud.wait_ops import significant_now as _wait_significant_now
 from cloud.wait_ops import get_handoff as _handoff_get
@@ -185,7 +184,6 @@ from cloud.sub2api import ensure_gateway_keys_table as _ensure_gateway_keys
 from cloud.sub2api import ensure_user_gateway_key as _ensure_sub2api_key
 from cloud.sub2api import gateway_name as _gateway_name
 from cloud.sub2api import gateway_origin as _gateway_origin
-from cloud.sub2api import get_user_gateway_key as _get_sub2api_key
 from cloud.sub2api import list_gateway_models as _list_sub2api_models
 from cloud.sub2api import rotate_user_gateway_key as _rotate_sub2api_key
 from cloud.sub2api import validate_or_rotate_gateway_key as _validate_sub2api_key
@@ -1822,14 +1820,6 @@ def _auth_store(db_path):
             pass
 
 
-def _row_get(row, name):
-    """sqlite Row 的容错取值：老库行缺列时返回 None 不炸。"""
-    try:
-        return row[name]
-    except (KeyError, IndexError, TypeError):
-        return None
-
-
 def _agent_tools_from_row(row):
     """agent_tools 列的容错解析：唯一实现在
     agent_def.parse_agent_tools_column（9.5-code 统一）；本
@@ -1846,6 +1836,14 @@ def _agent_tool_approvals_from_row(row):
         _row_get(row, "agent_tool_approvals"))
 
 
+# 注意：此处曾有一份与 cloud.peer_messages.row_get 逐行相同的本地
+# _row_get（try return row[name] except KeyError/IndexError/TypeError
+# -> None），它遮蔽了文件顶部的同名 import（F811 + mypy no-redef）。
+# 两份实现语义完全相同，删本地定义后全部调用点解析到 import 的
+# 同一份实现，行为不变；见 test_server_row_get_unified 回归测试。
+# 演练注：main 侧在此处新增的 _agent_tools_from_row /
+# _agent_tool_approvals_from_row 予以保留（它们调用的 _row_get
+# 现解析到 import，与本地定义逐行同义）。
 def _task_to_dict(row, attachments=None, total_tokens=None):
     depends_on = row["depends_on"]
     try:

cloud/skill_seed.py

@@ -42,7 +42,6 @@ idempotently, without clobbering tenant edits, and that
 skills/list sees them there.
 """
 
-import os
 import shutil
 from pathlib import Path
 

cloud/sub2api.py

@@ -120,7 +120,7 @@ def get_user_gateway_key(conn, user_id):
 # 每次新建实例都要登录一次（实测约 111ms，占开通链路约 1/3），
 # 共享后同一进程内第二次起免登录。401/刷新失败会回落到重登录，
 # 密码轮换后下一次登录即用新密码刷新缓存，不会卡死。
-_SHARED_ADMIN = {}
+_SHARED_ADMIN: dict[tuple[str, str], tuple[str | None, str | None, float]] = {}
 _SHARED_ADMIN_LOCK = threading.RLock()
 
 
@@ -201,7 +201,7 @@ def _call_with_login_retry(fn):
 # Redis 节流，控制面目前没有 Redis 客户端，未做）。
 _LOGIN_BUDGET_PER_MINUTE = 15
 _LOGIN_WINDOW_SECONDS = 60.0
-_login_slots = collections.deque()
+_login_slots: collections.deque[float] = collections.deque()
 _login_slots_lock = threading.Lock()
 
 

cloud/tests/test_automations.py

@@ -9,7 +9,6 @@ import os
 import sqlite3
 import sys
 import tempfile
-import time
 import unittest
 from pathlib import Path
 

cloud/tests/test_chatgpt_codex_skills.py

@@ -5,7 +5,6 @@
 frontmatter、名字与目录一致、正文非空、不含密钥占位。
 """
 
-import os
 import re
 import unittest
 from pathlib import Path

cloud/tests/test_cloud_dispatch_pr.py

@@ -16,7 +16,6 @@ import time
 import unittest
 from pathlib import Path
 from types import SimpleNamespace
-from unittest import mock
 
 try:
     from cloud.auth import AuthStore
@@ -25,7 +24,6 @@ try:
     from cloud.server import create_server, init_db
     from cloud.worker import Worker
     from cloud.workers import (
-        heartbeat_worker,
         list_workers,
         register_worker,
     )
@@ -37,7 +35,6 @@ except ImportError:
     from cloud.server import create_server, init_db
     from cloud.worker import Worker
     from cloud.workers import (
-        heartbeat_worker,
         list_workers,
         register_worker,
     )

cloud/tests/test_docker_sandbox.py

@@ -169,7 +169,8 @@ class DockerCommandTestCase(unittest.TestCase):
             ex = self._executor(seen)
             dispatched = ex.dispatch_task(
                 {"id": "t1", "prompt": "do it", "worktree": "/tmp/wt"})
-            ran = ex.run_task(
+            # run_task 调用本身有副作用（seen 记命令）：只去绑定，不删调用。
+            ex.run_task(
                 {"id": "t1", "prompt": "again", "worktree": "/tmp/wt"})
         self.assertEqual(dispatched["status"], "completed")
         self.assertEqual(dispatched["executor"], "docker")

cloud/tests/test_fallback_retry.py

@@ -22,13 +22,11 @@ from pathlib import Path
 from types import SimpleNamespace
 
 try:
-    from cloud.cost import CostStore
     from cloud.models import ModelRouter
     from cloud.server import init_db
     from cloud.worker import Worker
 except ImportError:  # discover -s cloud/tests 时把项目根目录补进 sys.path
     sys.path.insert(0, str(Path(__file__).resolve().parents[2]))
-    from cloud.cost import CostStore
     from cloud.models import ModelRouter
     from cloud.server import init_db
     from cloud.worker import Worker

cloud/tests/test_harness2.py

@@ -7,30 +7,26 @@
 账本：100.25, -20.10, 5.00 —— 带符号和 85.15，只加正数是 105.25。
 """
 
-import csv
 import subprocess
 import sys
 import tempfile
 import unittest
 from pathlib import Path
-from types import SimpleNamespace
 
 try:
-    from cloud import harness2
     from cloud.harness2 import (
-        Library, Rollout, EditResult, Budget, build_context, validate,
+        Library, Rollout, EditResult, validate,
         read_tree, write_tree, is_allowed, normalize_rel, autotrigger,
-        skill_names, diff_summary, diversity, improve_task, select_final,
+        diff_summary, diversity, improve_task, select_final,
-        assign_ids, has_experience,
+        assign_ids,
     )
 except ImportError:  # discover -s cloud/tests 时把项目根目录补进 sys.path
     sys.path.insert(0, str(Path(__file__).resolve().parents[2]))
-    from cloud import harness2
     from cloud.harness2 import (
-        Library, Rollout, EditResult, Budget, build_context, validate,
+        Library, Rollout, EditResult, validate,
         read_tree, write_tree, is_allowed, normalize_rel, autotrigger,
-        skill_names, diff_summary, diversity, improve_task, select_final,
+        diff_summary, diversity, improve_task, select_final,
-        assign_ids, has_experience,
+        assign_ids,
     )
 
 BASE_SCRIPT = """import csv

cloud/tests/test_harness2_tasks.py

@@ -6,7 +6,6 @@
 账本与 base 树同 test_harness2.py：带符号和 85.15，只加正数 105.25。
 """
 
-import csv
 import json
 import subprocess
 import sys
@@ -17,26 +16,24 @@ from http.server import BaseHTTPRequestHandler, ThreadingHTTPServer
 from pathlib import Path
 
 try:
-    from cloud import harness2_tasks
     from cloud.harness2 import (
         parse_choice, parse_deliverable_files, parse_experience,
         parse_harness_files, parse_no_change, parse_notes, parse_why,
     )
     from cloud.harness2_tasks import (
-        CodexEditor, CodexProbeAdapter, HttpTaskClient, TaskState,
+        CodexEditor, HttpTaskClient, TaskState,
         build_multipart, preflight_checks, run_improvement_via_tasks, run_sweep,
         run_task, unwrap_result_text, validate_run_tag,
     )
     from cloud.uploads import parse_multipart, parse_paths_field
 except ImportError:  # discover -s cloud/tests 时把项目根目录补进 sys.path
     sys.path.insert(0, str(Path(__file__).resolve().parents[2]))
-    from cloud import harness2_tasks
     from cloud.harness2 import (
         parse_choice, parse_deliverable_files, parse_experience,
         parse_harness_files, parse_no_change, parse_notes, parse_why,
     )
     from cloud.harness2_tasks import (
-        CodexEditor, CodexProbeAdapter, HttpTaskClient, TaskState,
+        CodexEditor, HttpTaskClient, TaskState,
         build_multipart, preflight_checks, run_improvement_via_tasks, run_sweep,
         run_task, unwrap_result_text, validate_run_tag,
     )

cloud/tests/test_mcp_coverage.py

@@ -0,0 +1,355 @@
+"""MCP 三模块覆盖补点（CI 第二阶段覆盖率）。
+
+- websearch/imagegen：handle_request 全分支（协议方法、工具
+  分发、未知工具、工具异常、非对象请求、通知）+ main() 的
+  stdio 循环（坏 JSON、通知不回包），工具执行用注入的
+  fake，不联网、不调真实模型/网关。
+- playwright：全部 14 个工具分发 + 未知工具 + 工具异常 +
+  协议方法 + _handle_line 的解析错误与 handler 异常分支。
+  fake browser 只返回标记 dict，断言分发到了正确方法。
+- flat-copy：三个模块 + 后端文件拷到隔离临时目录，在
+  PYTHONPATH 只含该目录的子进程里 import——cloud 包不可见，
+  必须走 flat-copy 回落；断言后端类的 __module__ 是顶层
+  模块名（不是 cloud.*），即回落真实发生（importlib 重构
+  的回归证据，见 F401/mypy no-redef 一节）。
+"""
+
+import io
+import json
+import os
+import shutil
+import subprocess
+import sys
+import tempfile
+import unittest
+from pathlib import Path
+from unittest import mock
+
+try:
+    from cloud.websearch_mcp import handle_request as ws_handle
+    from cloud.websearch_mcp import main as ws_main
+    from cloud.imagegen_mcp import handle_request as ig_handle
+    from cloud.imagegen_mcp import main as ig_main
+    from cloud import playwright_mcp as pw_mod
+    from cloud.playwright_mcp import handle_request as pw_handle
+    from cloud.playwright_mcp import _handle_line
+except ImportError:
+    sys.path.insert(0, str(Path(__file__).resolve().parents[2]))
+    from cloud.websearch_mcp import handle_request as ws_handle
+    from cloud.websearch_mcp import main as ws_main
+    from cloud.imagegen_mcp import handle_request as ig_handle
+    from cloud.imagegen_mcp import main as ig_main
+    from cloud import playwright_mcp as pw_mod
+    from cloud.playwright_mcp import handle_request as pw_handle
+    from cloud.playwright_mcp import _handle_line
+
+
+def _call(handle, method, req_id=1, name=None, args=None):
+    req = {"jsonrpc": "2.0", "id": req_id, "method": method}
+    if name is not None:
+        req["params"] = {"name": name, "arguments": args or {}}
+    return handle(req)
+
+
+class WebSearchMcpTest(unittest.TestCase):
+    def test_protocol_methods(self):
+        self.assertIsNone(ws_handle({"method": "notifications/initialized"}))
+        self.assertIsNone(ws_handle({"method": "ping"}))
+        ping = _call(ws_handle, "ping")
+        self.assertEqual(ping["result"], {})
+        init = {"jsonrpc": "2.0", "id": 1, "method": "initialize",
+                "params": {"protocolVersion": "2025-01-01"}}
+        resp = ws_handle(init)
+        self.assertEqual(resp["result"]["protocolVersion"], "2025-01-01")
+        self.assertEqual(resp["result"]["serverInfo"]["name"], "websearch")
+        self.assertIsNone(ws_handle({"method": "initialize"}))
+        listed = _call(ws_handle, "tools/list")
+        self.assertIn("web_search",
+                      [t["name"] for t in listed["result"]["tools"]])
+
+    def test_tools_call_and_errors(self):
+        fake_search = lambda q, limit=5, provider="auto": {  # noqa: E731
+            "results": [{"title": q}]}
+        # 工具执行一律用注入的 fake，不调真实 search（不联网）。
+        resp = ws_handle({"jsonrpc": "2.0", "id": 2, "method": "tools/call",
+                          "params": {"name": "web_search",
+                                     "arguments": {"query": "hello"}}},
+                         search_fn=fake_search)
+        self.assertFalse(resp["result"]["isError"])
+        self.assertIn("hello", resp["result"]["content"][0]["text"])
+        resp = ws_handle({"jsonrpc": "2.0", "id": 3, "method": "tools/call",
+                          "params": {"name": "web_fetch",
+                                     "arguments": {"url": "https://x/"}}},
+                         fetch_fn=lambda u, char_limit=15000: {"text": "T"})
+        self.assertIn("T", resp["result"]["content"][0]["text"])
+        unknown = _call(ws_handle, "tools/call", name="nope")
+        self.assertEqual(unknown["error"]["code"], -32602)
+        boom = ws_handle({"jsonrpc": "2.0", "id": 4, "method": "tools/call",
+                          "params": {"name": "web_search",
+                                     "arguments": {"query": "q"}}},
+                         search_fn=lambda *a, **k: (_ for _ in ()).throw(
+                             RuntimeError("boom")))
+        self.assertTrue(boom["result"]["isError"])
+        self.assertIn("boom", boom["result"]["content"][0]["text"])
+        self.assertEqual(ws_handle("nope")["error"]["code"], -32700)
+        self.assertEqual(
+            _call(ws_handle, "bogus/method")["error"]["code"], -32601)
+        self.assertIsNone(ws_handle(
+            {"method": "tools/call",
+             "params": {"name": "web_search", "arguments": {}}}))
+
+    def test_main_stdio_loop(self):
+        lines = ('{"bad json\n'
+                 '{"jsonrpc": "2.0", "id": 1, "method": "initialize",'
+                 ' "params": {}}\n'
+                 '{"jsonrpc": "2.0", "id": 2, "method": "tools/list"}\n'
+                 '{"jsonrpc": "2.0", "id": 3, "method": "ping"}\n'
+                 '{"jsonrpc": "2.0", "method": "notifications/initialized"}\n'
+                 '\n'
+                 '{"jsonrpc": "2.0", "id": 4, "method": "bogus"}\n')
+        out = io.StringIO()
+        with mock.patch.object(sys, "stdin", io.StringIO(lines)), \
+                mock.patch.object(sys, "stdout", out):
+            ws_main()
+        rows = [json.loads(x) for x in out.getvalue().splitlines()]
+        self.assertEqual(len(rows), 5)
+        self.assertEqual(rows[0]["error"]["code"], -32700)
+        self.assertEqual(rows[1]["result"]["serverInfo"]["name"],
+                         "websearch")
+        self.assertEqual(rows[3]["result"], {})
+        self.assertEqual(rows[4]["error"]["code"], -32601)
+
+
+class ImageGenMcpTest(unittest.TestCase):
+    def test_tools_call_and_errors(self):
+        gen = lambda p, output_filename=None, size="1024x1024", model=None: {  # noqa: E731
+            "path": "/w/gen.png", "prompt": p}
+        edit = lambda p, inp, output_filename=None, size="1024x1024", model=None: {  # noqa: E731
+            "path": "/w/edit.png", "input": inp}
+        resp = ig_handle({"jsonrpc": "2.0", "id": 1, "method": "tools/call",
+                          "params": {"name": "image_generate",
+                                     "arguments": {"prompt": "cat"}}},
+                         generate_fn=gen)
+        self.assertIn("gen.png", resp["result"]["content"][0]["text"])
+        resp = ig_handle({"jsonrpc": "2.0", "id": 2, "method": "tools/call",
+                          "params": {"name": "image_edit",
+                                     "arguments": {"prompt": "dog",
+                                                   "input_filename": "in.png"}}},
+                         edit_fn=edit)
+        self.assertIn("in.png", resp["result"]["content"][0]["text"])
+        unknown = _call(ig_handle, "tools/call", name="nope")
+        self.assertEqual(unknown["error"]["code"], -32602)
+        boom = ig_handle({"jsonrpc": "2.0", "id": 3, "method": "tools/call",
+                          "params": {"name": "image_generate",
+                                     "arguments": {"prompt": "x"}}},
+                         generate_fn=lambda *a, **k: (_ for _ in ()).throw(
+                             RuntimeError("gen boom")))
+        self.assertTrue(boom["result"]["isError"])
+        self.assertIn("gen boom", boom["result"]["content"][0]["text"])
+        self.assertEqual(ig_handle(42)["error"]["code"], -32700)
+        self.assertIsNone(ig_handle({"method": "ping"}))
+        self.assertEqual(_call(ig_handle, "ping")["result"], {})
+
+    def test_main_stdio_loop(self):
+        lines = ('{"jsonrpc": "2.0", "id": 1, "method": "initialize",'
+                 ' "params": {}}\n'
+                 '{"jsonrpc": "2.0", "id": 2, "method": "tools/list"}\n'
+                 'not json\n')
+        out = io.StringIO()
+        with mock.patch.object(sys, "stdin", io.StringIO(lines)), \
+                mock.patch.object(sys, "stdout", out):
+            ig_main()
+        rows = [json.loads(x) for x in out.getvalue().splitlines()]
+        self.assertEqual(len(rows), 3)
+        self.assertEqual(rows[0]["result"]["serverInfo"]["name"],
+                         "imagegen")
+        self.assertEqual(rows[2]["error"]["code"], -32700)
+
+
+class FakeBrowserFull:
+    def __init__(self):
+        self.calls = []
+
+    def navigate(self, url):
+        self.calls.append(("navigate", url))
+        return {"tool": "navigate", "url": url}
+
+    def snapshot(self):
+        self.calls.append(("snapshot",))
+        return {"tool": "snapshot"}
+
+    def click(self, selector):
+        self.calls.append(("click", selector))
+        return {"tool": "click"}
+
+    def fill(self, selector, text):
+        self.calls.append(("fill", selector, text))
+        return {"tool": "fill"}
+
+    def click_at(self, x, y):
+        self.calls.append(("click_at", x, y))
+        return {"tool": "click_at"}
+
+    def mouse_move(self, x, y):
+        self.calls.append(("mouse_move", x, y))
+        return {"tool": "mouse_move"}
+
+    def key_press(self, key):
+        self.calls.append(("key_press", key))
+        return {"tool": "key_press"}
+
+    def type_text(self, text):
+        self.calls.append(("type_text", text))
+        return {"tool": "type_text"}
+
+    def snapshot_refs(self):
+        self.calls.append(("snapshot_refs",))
+        return {"tool": "snapshot_refs"}
+
+    def click_ref(self, ref):
+        self.calls.append(("click_ref", ref))
+        return {"tool": "click_ref"}
+
+    def live_shot(self, name=None):
+        self.calls.append(("live_shot", name))
+        return {"tool": "live_shot"}
+
+    def evaluate(self, expression):
+        self.calls.append(("evaluate", expression))
+        return {"tool": "evaluate"}
+
+    def screenshot(self, name=None):
+        self.calls.append(("screenshot", name))
+        return {"tool": "screenshot"}
+
+    def close(self):
+        self.calls.append(("close",))
+        return {"tool": "close"}
+
+
+class PlaywrightMcpTest(unittest.TestCase):
+    def test_all_tools_dispatch(self):
+        cases = [
+            ("browser_navigate", {"url": "https://x/"}, "navigate"),
+            ("browser_snapshot", {}, "snapshot"),
+            ("browser_click", {"selector": "#a"}, "click"),
+            ("browser_fill", {"selector": "#a", "text": "hi"}, "fill"),
+            ("browser_click_at", {"x": 1, "y": 2}, "click_at"),
+            ("browser_mouse_move", {"x": 1, "y": 2}, "mouse_move"),
+            ("browser_key_press", {"key": "Enter"}, "key_press"),
+            ("browser_type_text", {"text": "hi"}, "type_text"),
+            ("browser_snapshot_refs", {}, "snapshot_refs"),
+            ("browser_click_ref", {"ref": "e0"}, "click_ref"),
+            ("browser_live_shot", {}, "live_shot"),
+            ("browser_evaluate", {"expression": "1+1"}, "evaluate"),
+            ("browser_screenshot", {}, "screenshot"),
+            ("browser_close", {}, "close"),
+        ]
+        for i, (name, args, tool) in enumerate(cases):
+            fake = FakeBrowserFull()
+            resp = pw_handle(
+                {"jsonrpc": "2.0", "id": i + 1, "method": "tools/call",
+                 "params": {"name": name, "arguments": args}},
+                browser=fake)
+            self.assertFalse(resp["result"]["isError"], name)
+            self.assertIn(tool, resp["result"]["content"][0]["text"], name)
+
+    def test_unknown_tool_and_tool_exception(self):
+        fake = FakeBrowserFull()
+        unknown = pw_handle(
+            {"jsonrpc": "2.0", "id": 1, "method": "tools/call",
+             "params": {"name": "browser_nope", "arguments": {}}},
+            browser=fake)
+        self.assertEqual(unknown["error"]["code"], -32602)
+
+        class Raising(FakeBrowserFull):
+            def navigate(self, url):
+                raise RuntimeError("nav boom")
+
+        resp = pw_handle(
+            {"jsonrpc": "2.0", "id": 2, "method": "tools/call",
+             "params": {"name": "browser_navigate",
+                        "arguments": {"url": "https://x/"}}},
+            browser=Raising())
+        self.assertTrue(resp["result"]["isError"])
+        self.assertIn("nav boom", resp["result"]["content"][0]["text"])
+
+    def test_protocol_and_handle_line(self):
+        self.assertEqual(pw_handle("x")["error"]["code"], -32700)
+        self.assertIsNone(pw_handle({"method": "ping"}))
+        self.assertEqual(_call(pw_handle, "ping")["result"], {})
+        out = io.StringIO()
+        with mock.patch.object(sys, "stdout", out):
+            _handle_line('{"jsonrpc": "2.0", "id": 9, "method": "ping"}',
+                         FakeBrowserFull())
+            _handle_line("{bad json", FakeBrowserFull())
+        rows = [json.loads(x) for x in out.getvalue().splitlines()]
+        self.assertEqual(rows[0]["result"], {})
+        self.assertEqual(rows[1]["error"]["code"], -32700)
+        with mock.patch.object(pw_mod, "handle_request",
+                               side_effect=RuntimeError("handler boom")):
+            out2, err2 = io.StringIO(), io.StringIO()
+            with mock.patch.object(sys, "stdout", out2), \
+                    mock.patch.object(sys, "stderr", err2):
+                _handle_line('{"jsonrpc": "2.0", "id": 10, '
+                             '"method": "tools/list"}',
+                             FakeBrowserFull())
+        row = json.loads(out2.getvalue().strip())
+        self.assertEqual(row["error"]["code"], -32603)
+        self.assertIn("handler boom", err2.getvalue())
+
+
+class FlatCopyTest(unittest.TestCase):
+    def _flat_copy(self, pairs):
+        td = tempfile.mkdtemp()
+        self.addCleanup(shutil.rmtree, td, True)
+        root = Path(__file__).resolve().parents[2] / "cloud"
+        for name in pairs:
+            shutil.copy(root / name, os.path.join(td, name))
+        return td
+
+    def _run_flat(self, td, module, check):
+        script = ("import %s as m; print(m.SERVER_NAME); print(check(m))"
+                  % module)
+        env = {"PATH": os.environ.get("PATH", ""), "PYTHONPATH": td}
+        proc = subprocess.run(
+            [sys.executable, "-c",
+             script.replace("check(m)", check)],
+            cwd=td, env=env, capture_output=True, text=True, timeout=60)
+        self.assertEqual(proc.returncode, 0, proc.stderr)
+        return proc.stdout
+
+    def test_playwright_flat_copy(self):
+        td = self._flat_copy(["playwright_mcp.py",
+                              "playwright_browser.py"])
+        out = self._run_flat(
+            td, "playwright_mcp",
+            "m.Browser.__module__ + '|' + "
+            "str(m.handle_request({'jsonrpc': '2.0', 'id': 1,"
+            " 'method': 'tools/list'})['result']['tools'][0]['name'])")
+        self.assertIn("playwright", out)
+        self.assertIn("playwright_browser|browser_navigate", out)
+
+    def test_websearch_flat_copy(self):
+        td = self._flat_copy(["websearch_mcp.py", "websearch.py"])
+        out = self._run_flat(
+            td, "websearch_mcp",
+            "m.search.__module__ + '|' + "
+            "str(m.handle_request({'jsonrpc': '2.0', 'id': 1,"
+            " 'method': 'ping'})['result'])")
+        self.assertIn("websearch", out)
+        self.assertIn("websearch|{}", out)
+
+    def test_imagegen_flat_copy(self):
+        td = self._flat_copy(["imagegen_mcp.py", "imagegen.py"])
+        out = self._run_flat(
+            td, "imagegen_mcp",
+            "m.generate.__module__ + '|' + "
+            "str(m.handle_request({'jsonrpc': '2.0', 'id': 1,"
+            " 'method': 'ping'})['result'])")
+        self.assertIn("imagegen", out)
+        self.assertIn("imagegen|{}", out)
+
+
+if __name__ == "__main__":
+    unittest.main()

cloud/tests/test_migrate_cli.py

@@ -0,0 +1,75 @@
+"""cloud/migrate_sqlite_to_postgres.py CLI 包装层单测
+（CI 第二阶段覆盖率：基线 0%）。
+
+只测包装层本身：参数解析、行数合计打印、missing 警告分支、
+unmigrated/skipped 报告行。真正的迁移逻辑在
+cloud/migrations.migrate_sqlite_to_postgres，另有
+test_p2_migrations / test_postgres_integration 覆盖；
+本文件 patch 的是包装层调用的两个函数名，不碰迁移本身。
+"""
+
+import io
+import sys
+import unittest
+from contextlib import redirect_stdout
+from pathlib import Path
+from unittest import mock
+
+try:
+    from cloud import migrate_sqlite_to_postgres as mig_cli
+except ImportError:
+    sys.path.insert(0, str(Path(__file__).resolve().parents[2]))
+    from cloud import migrate_sqlite_to_postgres as mig_cli
+
+
+def _coverage(**over):
+    base = {"missing_from_migration": [],
+            "unmigrated_no_pg_schema": ["legacy_x"],
+            "skipped_by_design": ["audit_log"]}
+    base.update(over)
+    return base
+
+
+class MigrateCliTest(unittest.TestCase):
+    def test_totals_and_reports(self):
+        with mock.patch.object(
+                mig_cli, "migrate_sqlite_to_postgres",
+                return_value={"users": 2, "tasks": 0}), \
+                mock.patch.object(mig_cli, "migration_coverage",
+                                  return_value=_coverage()):
+            out = io.StringIO()
+            with redirect_stdout(out):
+                rc = mig_cli.main(["--sqlite", "/tmp/x.db"])
+        self.assertEqual(rc, 0)
+        text = out.getvalue()
+        self.assertIn("users: 2", text)
+        self.assertIn("tasks: 0", text)
+        self.assertIn("total inserted: 2", text)
+        self.assertNotIn("WARNING", text)
+        self.assertIn("unmigrated (no PG schema, NOT migrated): legacy_x",
+                      text)
+        self.assertIn("skipped by design: audit_log", text)
+
+    def test_missing_warning_branch(self):
+        with mock.patch.object(
+                mig_cli, "migrate_sqlite_to_postgres",
+                return_value={}), \
+                mock.patch.object(
+                    mig_cli, "migration_coverage",
+                    return_value=_coverage(
+                        missing_from_migration=["tasks"])):
+            out = io.StringIO()
+            with redirect_stdout(out):
+                rc = mig_cli.main(["--sqlite", "/tmp/x.db"])
+        self.assertEqual(rc, 0)
+        self.assertIn("WARNING missing_from_migration: tasks",
+                      out.getvalue())
+
+    def test_sqlite_arg_required(self):
+        with self.assertRaises(SystemExit) as ctx:
+            mig_cli.main([])
+        self.assertEqual(ctx.exception.code, 2)
+
+
+if __name__ == "__main__":
+    unittest.main()

cloud/tests/test_oauth.py

@@ -6,7 +6,6 @@ GitHub / Google 发任何请求；出现的 client_id / secret 哨兵值
 """
 
 import os
-import sqlite3
 import sys
 import tempfile
 import time

cloud/tests/test_pr_jobs.py

@@ -19,7 +19,6 @@ import tempfile
 import threading
 import unittest
 from pathlib import Path
-from types import SimpleNamespace
 
 try:
     from cloud.auth import AuthStore
@@ -84,7 +83,8 @@ class PrJobsTableTestCase(unittest.TestCase):
         conn = _db_connect(self.db_path)
         try:
             job1 = enqueue_pr_job(conn, "t1", "p1", "host-a")
-            job2 = enqueue_pr_job(conn, "t2", "p1", "host-a")
+            # 第二条入队本身是测试前置（claimed2 要拿到 t2）：只去绑定，不删调用。
+            enqueue_pr_job(conn, "t2", "p1", "host-a")
             self.assertIsNotNone(find_open_pr_job(conn, "t1"))
             claimed = claim_pr_job(conn, "host-a")
             self.assertEqual(claimed["task_id"], "t1")

cloud/tests/test_prod_worker_timeout.py

@@ -17,13 +17,11 @@ from unittest import mock
 
 try:
     from cloud import backup as backup_mod
-    from cloud.auth import AuthStore
     from cloud.sandbox import DockerExecutor
     from cloud.worker import Worker
 except ImportError:
     sys.path.insert(0, str(Path(__file__).resolve().parents[2]))
     from cloud import backup as backup_mod
-    from cloud.auth import AuthStore
     from cloud.sandbox import DockerExecutor
     from cloud.worker import Worker
 

cloud/tests/test_server_goal_notify.py

@@ -22,12 +22,12 @@ from types import SimpleNamespace
 try:
     from cloud.auth import AuthStore
     from cloud.push import create_subscription
-    from cloud.server import create_server, init_db
+    from cloud.server import create_server
 except ImportError:
     sys.path.insert(0, str(Path(__file__).resolve().parents[2]))
     from cloud.auth import AuthStore
     from cloud.push import create_subscription
-    from cloud.server import create_server, init_db
+    from cloud.server import create_server
 
 
 def _result(returncode=0, stdout="", stderr=""):

cloud/tests/test_server_row_get.py

@@ -0,0 +1,44 @@
+"""server._row_get 统一回归测试（CI 第二阶段 F811/mypy no-redef）。
+
+背景：cloud/server.py 顶部 import cloud.peer_messages.row_get
+为 _row_get，文件后部又定义了一份逐行相同的本地 _row_get
+（try return row[name] except KeyError/IndexError/TypeError
+-> None），遮蔽 import。本轨删本地定义，全部调用点解析到
+import 的同一份实现。本测试钉住两点：
+
+- server._row_get 与 peer_messages.row_get 是同一对象；
+- 缺列/空行/正常取值三路行为与原来本地定义一致。
+"""
+
+import sys
+import unittest
+from pathlib import Path
+
+try:
+    from cloud.peer_messages import row_get as peer_row_get
+    from cloud.server import _row_get
+except ImportError:
+    sys.path.insert(0, str(Path(__file__).resolve().parents[2]))
+    from cloud.peer_messages import row_get as peer_row_get
+    from cloud.server import _row_get
+
+
+class RowGetUnifiedTest(unittest.TestCase):
+    def test_same_implementation(self):
+        self.assertIs(_row_get, peer_row_get)
+
+    def test_present_and_missing(self):
+        self.assertEqual(_row_get({"a": 1}, "a"), 1)
+        self.assertIsNone(_row_get({"a": 1}, "missing"))
+        self.assertIsNone(_row_get(None, "a"))
+
+    def test_index_error_row(self):
+        class OldRow:
+            def __getitem__(self, name):
+                raise IndexError(name)
+
+        self.assertIsNone(_row_get(OldRow(), "missing_col"))
+
+
+if __name__ == "__main__":
+    unittest.main()

cloud/tests/test_sub2api.py

@@ -494,7 +494,6 @@ class TestDockerSecretEnv(unittest.TestCase):
         return ex
 
     def test_secret_name_only_in_argv(self):
-        from cloud.sandbox import DockerExecutor
         seen = {}
 
         def runner(cmd, timeout=None, env=None):
@@ -517,7 +516,6 @@ class TestDockerSecretEnv(unittest.TestCase):
         self.assertIsNone(ex._docker_cli_env(None))
 
     def test_call_runner_env_loud_fail(self):
-        from cloud.sandbox import DockerExecutor
 
         def cmd_only(cmd, timeout=None):
             return object()

cloud/tests/test_task_environments.py

@@ -18,7 +18,6 @@ import tempfile
 import threading
 import time
 import unittest
-import uuid
 from pathlib import Path
 from types import SimpleNamespace
 

cloud/tests/test_task_links.py

@@ -19,13 +19,13 @@ from types import SimpleNamespace
 
 try:
     from cloud.auth import AuthStore
-    from cloud.sandbox import DockerExecutor, LocalExecutor
+    from cloud.sandbox import DockerExecutor
     from cloud.server import create_server, init_db
     from cloud.worker import Worker
 except ImportError:  # discover -s cloud/tests 时把项目根目录补进 sys.path
     sys.path.insert(0, str(Path(__file__).resolve().parents[2]))
     from cloud.auth import AuthStore
-    from cloud.sandbox import DockerExecutor, LocalExecutor
+    from cloud.sandbox import DockerExecutor
     from cloud.server import create_server, init_db
     from cloud.worker import Worker
 

cloud/tests/test_task_trace.py

@@ -8,7 +8,6 @@
 import http.client
 import json
 import os
-import socket
 import sqlite3
 import sys
 import tempfile

cloud/tests/test_webhooks.py

@@ -19,7 +19,6 @@ sys.path.insert(
     0, os.path.normpath(os.path.join(os.path.dirname(__file__), ".."))
 )
 
-import webhooks  # noqa: E402
 from webhooks import (  # noqa: E402
     ALLOWED_NOTIFICATION_EVENTS,
     NotificationService,

cloud/wait_ops.py

@@ -36,7 +36,6 @@ automations 同款），不在 POSTGRES_SCHEMA 里，因此必须同步
 
 import base64
 import json
-import os
 import re
 import time
 import uuid

cloud/webhooks.py

@@ -18,7 +18,6 @@
 
 import hashlib
 import hmac
-import sqlite3
 import time
 
 from cloud.db import connect as _db_connect

cloud/websearch.py

@@ -58,8 +58,6 @@ import html as _html
 import ipaddress
 import json
 import os
-import re
-import socket
 import urllib.error
 import urllib.parse
 import urllib.request

cloud/websearch_mcp.py

@@ -15,15 +15,23 @@ copying; it never writes API keys into config.toml -- keys are read
 from the MCP server process environment by websearch.py.
 """
 
+import importlib
 import json
 import os
 import sys
 
+# cloud 包优先、租户 flat copy 回落（与 playwright_mcp 同款理由：
+# 两次 from-import 会让 mypy 报 no-redef，importlib 属性绑定
+# 在两条路径下都是同一组名字；cloud 导入本身的 ImportError
+# 同样回落，与原来 except 语义一致）。
 try:
-    from cloud.websearch import WebSearchError, fetch_best, search
+    _mcp_backend = importlib.import_module("cloud.websearch")
 except ImportError:  # flat copy in a tenant codex-home
     sys.path.insert(0, os.path.dirname(os.path.abspath(__file__)))
-    from websearch import WebSearchError, fetch_best, search  # noqa: E402
+    _mcp_backend = importlib.import_module("websearch")
+WebSearchError = _mcp_backend.WebSearchError
+fetch_best = _mcp_backend.fetch_best
+search = _mcp_backend.search
 
 SERVER_NAME = "websearch"
 SERVER_VERSION = "0.1.0"

deploy/deploy.py

@@ -44,7 +44,6 @@ import hashlib
 import json
 import os
 import re
-import shutil
 import subprocess
 import sys
 import tarfile

deploy/tests/test_deploy.py

@@ -46,7 +46,8 @@ class TestDeployPure(unittest.TestCase):
         with tempfile.NamedTemporaryFile(suffix=".tar.gz", delete=False) as f:
             path = f.name
         try:
-            with tarfile.open(path, "w:gz") as tf:
+            # 空 tarball 文件本身是测试前置（tarball_commit 应拒绝）：上下文照跑，只去 as 绑定。
+            with tarfile.open(path, "w:gz"):
                 pass
             with self.assertRaises(SystemExit):
                 deploy_mod.tarball_commit(path)

deploy/tests/test_deploy_gates.py

@@ -0,0 +1,148 @@
+"""deploy.py 门禁编排单测（CI 第二阶段覆盖率补点）。
+
+只测编排层：run() 的 check 开关、check_clean_tree 的
+拒绝/告警/放行、deploy_via_git 的 dry-run 与各门禁拒绝、
+main() 的 git 分发与 tar 路径前置检查。真实 SSH/传输仍由
+test_deploy.py 的 transfer/remote_sha 测试覆盖，本文件不
+碰网络与主机。
+"""
+
+import os
+import subprocess
+import sys
+import unittest
+from types import SimpleNamespace
+from unittest import mock
+
+REPO_ROOT = os.path.join(os.path.dirname(__file__), "..", "..")
+sys.path.insert(0, os.path.join(REPO_ROOT, "deploy"))
+import deploy as deploy_mod  # noqa: E402
+
+
+def _args(**over):
+    base = dict(tarball=None, build_only=False, host=None, user="root",
+                dry_run=True, repo="git@github.com:openqj/codex-app-server",
+                git_key="/tmp/k", allow_dirty=False, skip_tests=False,
+                ref="HEAD", commit=None, source="git",
+                out_dir="/tmp/out")
+    base.update(over)
+    return SimpleNamespace(**base)
+
+
+class RunAndCleanTreeTest(unittest.TestCase):
+    def test_run_check_exits_on_failure(self):
+        bad = subprocess.CompletedProcess(args=[], returncode=1,
+                                          stdout="", stderr="boom")
+        with mock.patch.object(deploy_mod.subprocess, "run",
+                               return_value=bad):
+            with self.assertRaises(SystemExit) as ctx:
+                deploy_mod.run(["false"])
+            self.assertEqual(ctx.exception.code, 1)
+
+    def test_run_no_check_returns_result(self):
+        bad = subprocess.CompletedProcess(args=[], returncode=1,
+                                          stdout="o", stderr="e")
+        with mock.patch.object(deploy_mod.subprocess, "run",
+                               return_value=bad):
+            self.assertIs(deploy_mod.run(["false"], check=False), bad)
+
+    def test_clean_tree_dirty_refuses_and_warns(self):
+        dirty = SimpleNamespace(returncode=1)
+        with mock.patch.object(deploy_mod, "run", return_value=dirty):
+            with self.assertRaises(SystemExit) as ctx:
+                deploy_mod.check_clean_tree(False)
+            self.assertEqual(ctx.exception.code, 2)
+            deploy_mod.check_clean_tree(True)  # 只告警，不退出
+
+    def test_clean_tree_clean_passes(self):
+        clean = SimpleNamespace(returncode=0)
+        with mock.patch.object(deploy_mod, "run", return_value=clean):
+            deploy_mod.check_clean_tree(False)
+
+
+class DeployViaGitTest(unittest.TestCase):
+    def _patched(self, commit="a" * 40, on_main=True):
+        return (mock.patch.object(deploy_mod, "check_clean_tree"),
+                mock.patch.object(deploy_mod, "run_tests"),
+                mock.patch.object(deploy_mod, "git_commit",
+                                  return_value=commit),
+                mock.patch.object(deploy_mod, "commit_on_origin_main",
+                                  return_value=on_main))
+
+    def test_dry_run_success(self):
+        patches = self._patched()
+        for p in patches:
+            p.start()
+        try:
+            self.assertEqual(deploy_mod.deploy_via_git(_args()), 0)
+        finally:
+            for p in patches:
+                p.stop()
+
+    def test_gates(self):
+        cases = [
+            (_args(tarball="/tmp/x.tgz"), "tarball 冲突"),
+            (_args(repo="https://github.com/o/r"), "非 SSH 仓库"),
+        ]
+        for args, _why in cases:
+            patches = self._patched()
+            for p in patches:
+                p.start()
+            try:
+                with self.assertRaises(SystemExit) as ctx:
+                    deploy_mod.deploy_via_git(args)
+                self.assertEqual(ctx.exception.code, 2)
+            finally:
+                for p in patches:
+                    p.stop()
+        patches = self._patched(commit="abc")
+        for p in patches:
+            p.start()
+        try:
+            with self.assertRaises(SystemExit) as ctx:
+                deploy_mod.deploy_via_git(_args())
+            self.assertEqual(ctx.exception.code, 2)
+        finally:
+            for p in patches:
+                p.stop()
+        patches = self._patched(on_main=False)
+        for p in patches:
+            p.start()
+        try:
+            with self.assertRaises(SystemExit) as ctx:
+                deploy_mod.deploy_via_git(_args())
+            self.assertEqual(ctx.exception.code, 2)
+        finally:
+            for p in patches:
+                p.stop()
+
+
+class MainDispatchTest(unittest.TestCase):
+    def test_main_dispatches_git(self):
+        seen = []
+        with mock.patch.object(
+                deploy_mod, "deploy_via_git",
+                side_effect=lambda a: seen.append(a) or 0):
+            rc = deploy_mod.main(["h", "root", "--source", "git",
+                                  "--dry-run"])
+        self.assertEqual(rc, 0)
+        self.assertEqual(seen[0].source, "git")
+        self.assertTrue(seen[0].dry_run)
+
+    def test_main_tar_needs_host_or_mode(self):
+        with self.assertRaises(SystemExit) as ctx:
+            deploy_mod.main([])
+        self.assertEqual(ctx.exception.code, 2)
+
+    def test_main_build_only_tar_path(self):
+        with mock.patch.object(deploy_mod, "check_clean_tree"), \
+                mock.patch.object(deploy_mod, "run_tests"), \
+                mock.patch.object(
+                    deploy_mod, "build_release",
+                    return_value=("tb.tgz", "sha", "c" * 40)):
+            rc = deploy_mod.main(["myhost", "--build-only"])
+        self.assertEqual(rc, 0)
+
+
+if __name__ == "__main__":
+    unittest.main()

deploy/tests/test_host_git_fetch.py

@@ -6,7 +6,6 @@ host_deploy.sh 代替真实换目录脚本：假脚本只验 tar sha 与
 DEPLOYED_COMMIT，打印 RELEASE_OK，不碰 systemctl、不动 live。
 host_git_fetch.sh 的 CODEX_GIT_* 环境变量把工作区指到临时目录。
 """
-import hashlib
 import os
 import subprocess
 import tarfile

deploy/tests/test_verify_multiuser_helpers.py

@@ -0,0 +1,129 @@
+"""deploy/verify-multiuser.py 纯 helper 单测（CI 第二阶段覆盖率：
+基线 0%）。只测 check() 与 req()：
+
+- check() 写 RESULTS 台账并打印 PASS/FAIL；
+- req() 对本地回环 fake HTTP 成功/HTTPError/非 JSON 体/
+  连接拒绝四路的返回值与请求头。
+
+main() 是主机侧对真实控制面的多用户实弹验证（建
+multiuser-test-* 用户/项目/任务、调 merge），按任务门禁
+“不要建生产任务”，本轨不以 mock 冒充它——main() 仍为
+未覆盖，如实留在缺口里。
+"""
+
+import http.server
+import importlib.util
+import io
+import json
+import os
+import socket
+import threading
+import unittest
+from contextlib import redirect_stdout
+from pathlib import Path
+
+REPO_ROOT = Path(__file__).resolve().parents[2]
+_SPEC = importlib.util.spec_from_file_location(
+    "verify_multiuser",
+    os.path.join(REPO_ROOT, "deploy", "verify-multiuser.py"))
+vm = importlib.util.module_from_spec(_SPEC)
+_SPEC.loader.exec_module(vm)
+
+
+class Handler(http.server.BaseHTTPRequestHandler):
+    seen = []
+
+    def _send(self, code, body, ctype="application/json"):
+        data = body if isinstance(body, bytes) else body.encode()
+        self.send_response(code)
+        self.send_header("Content-Type", ctype)
+        self.send_header("Content-Length", str(len(data)))
+        self.end_headers()
+        self.wfile.write(data)
+
+    def do_GET(self):
+        Handler.seen.append((self.path, dict(self.headers)))
+        if self.path == "/ok":
+            self._send(200, json.dumps({"ok": True}))
+        elif self.path == "/nf":
+            self._send(404, json.dumps({"error": "nope"}))
+        else:
+            self._send(404, "not json", "text/plain")
+
+    def do_POST(self):
+        length = int(self.headers.get("Content-Length") or 0)
+        Handler.seen.append(
+            (self.path, dict(self.headers), self.rfile.read(length)))
+        self._send(201, json.dumps({"created": True}))
+
+    def log_message(self, *args):
+        pass
+
+
+class ReqTest(unittest.TestCase):
+    @classmethod
+    def setUpClass(cls):
+        cls.httpd = http.server.ThreadingHTTPServer(
+            ("127.0.0.1", 0), Handler)
+        cls.port = cls.httpd.server_address[1]
+        cls.thread = threading.Thread(target=cls.httpd.serve_forever,
+                                      daemon=True)
+        cls.thread.start()
+
+    @classmethod
+    def tearDownClass(cls):
+        cls.httpd.shutdown()
+        cls.httpd.server_close()
+
+    def setUp(self):
+        Handler.seen = []
+
+    def base(self):
+        return "http://127.0.0.1:%d" % self.port
+
+    def test_get_ok(self):
+        code, body = vm.req(self.base(), "GET", "/ok")
+        self.assertEqual((code, body), (200, {"ok": True}))
+
+    def test_http_error_json_and_non_json(self):
+        code, body = vm.req(self.base(), "GET", "/nf")
+        self.assertEqual((code, body), (404, {"error": "nope"}))
+        code, body = vm.req(self.base(), "GET", "/bad")
+        self.assertEqual((code, body), (404, {}))
+
+    def test_post_headers_and_body(self):
+        code, body = vm.req(self.base(), "POST", "/ok",
+                            key="k-123", body={"a": 1})
+        self.assertEqual(code, 201)
+        self.assertEqual(body, {"created": True})
+        path, headers, raw = Handler.seen[0]
+        self.assertEqual(headers.get("X-Api-Key"), "k-123")
+        self.assertEqual(headers.get("Content-Type"), "application/json")
+        self.assertEqual(json.loads(raw), {"a": 1})
+
+    def test_connection_refused_returns_none(self):
+        sock = socket.socket()
+        sock.bind(("127.0.0.1", 0))
+        dead_port = sock.getsockname()[1]
+        sock.close()
+        code, body = vm.req("http://127.0.0.1:%d" % dead_port,
+                            "GET", "/ok")
+        self.assertIsNone(code)
+        self.assertIn("error", body)
+
+
+class CheckTest(unittest.TestCase):
+    def test_check_appends_and_prints(self):
+        vm.RESULTS.clear()
+        out = io.StringIO()
+        with redirect_stdout(out):
+            vm.check("alpha", True)
+            vm.check("beta", 0)
+        self.assertEqual(vm.RESULTS, [("alpha", True), ("beta", False)])
+        self.assertIn("PASS alpha", out.getvalue())
+        self.assertIn("FAIL beta", out.getvalue())
+        vm.RESULTS.clear()
+
+
+if __name__ == "__main__":
+    unittest.main()

deploy/verify-multiuser.py

@@ -56,7 +56,7 @@ P1 = PREFIX + "p1"
 P2 = PREFIX + "p2"
 NEVER_DEP = PREFIX + "never-completes"
 
-RESULTS = []
+RESULTS: list[tuple[str, bool]] = []
 
 
 def check(name, ok):

examples/harness2_demo.py

@@ -12,7 +12,6 @@ base 会漏掉它）；确定性编辑器去掉正数过滤；选择器按产物
 """
 
 import argparse
-import csv
 import subprocess
 import sys
 import tempfile

host.py

@@ -1754,7 +1754,8 @@ def dispatch_from_branch(server, state, name, prompt, repo, dep_branch,
     核对把它拦下来回滚了）。"""
     repo = os.path.abspath(repo)
     git(["add", "-A"], dep_worktree)
-    c = git(["commit", "-m", f"host: {name} base from {dep_branch}"],
+    # 提交本身是依赖派发的前置（分支须含依赖改动）：只去绑定，不删调用。
+    git(["commit", "-m", f"host: {name} base from {dep_branch}"],
         dep_worktree)
     # nothing to commit 时分支里已有全部改动，继续用它当基底
     base = os.path.join(HERE, "work", "host-worktrees")
@@ -2304,16 +2305,22 @@ def main():
     p.add_argument("--max-retries", type=int, default=0)
 
     sub.add_parser("list")
-    p = sub.add_parser("status"); p.add_argument("--task", required=True)
+    p = sub.add_parser("status")
-    p = sub.add_parser("review"); p.add_argument("--task", required=True)
+    p.add_argument("--task", required=True)
-    p = sub.add_parser("merge"); p.add_argument("--task", required=True)
+    p = sub.add_parser("review")
+    p.add_argument("--task", required=True)
+    p = sub.add_parser("merge")
+    p.add_argument("--task", required=True)
     p.add_argument("--verify", default=None,
                    help="合并后验测命令；不传时仓库有 tests/ 就默认跑 unittest")
     p.add_argument("--no-verify", action="store_true")
-    p = sub.add_parser("review-comment"); p.add_argument("--task", required=True)
+    p = sub.add_parser("review-comment")
-    p.add_argument("--file", required=True); p.add_argument("--line", type=int, required=True)
+    p.add_argument("--task", required=True)
+    p.add_argument("--file", required=True)
+    p.add_argument("--line", type=int, required=True)
     p.add_argument("--comment", required=True)
-    p = sub.add_parser("fix-review"); p.add_argument("--task", required=True)
+    p = sub.add_parser("fix-review")
+    p.add_argument("--task", required=True)
     sub.add_parser("doctor")
     sub.add_parser("failures")
     sub.add_parser("health")
@@ -2328,35 +2335,61 @@ def main():
                    help="前台跑循环（测试用，不双 fork）")
     sub.add_parser("daemon-status")
     sub.add_parser("daemon-stop")
-    p = sub.add_parser("discard"); p.add_argument("--task", required=True)
+    p = sub.add_parser("discard")
-    p = sub.add_parser("archive"); p.add_argument("--task", required=True)
+    p.add_argument("--task", required=True)
-    p = sub.add_parser("fork"); p.add_argument("--task", required=True); p.add_argument("--name", required=True); p.add_argument("--sandbox", default=None)
+    p = sub.add_parser("archive")
-    p = sub.add_parser("compact"); p.add_argument("--task", required=True)
+    p.add_argument("--task", required=True)
-    p = sub.add_parser("code-review"); p.add_argument("--task", required=True)
+    p = sub.add_parser("fork")
+    p.add_argument("--task", required=True)
+    p.add_argument("--name", required=True)
+    p.add_argument("--sandbox", default=None)
+    p = sub.add_parser("compact")
+    p.add_argument("--task", required=True)
+    p = sub.add_parser("code-review")
+    p.add_argument("--task", required=True)
     p.add_argument("--base-branch", default=None,
                    help="评审目标：与该分支的差异（baseBranch）")
     p.add_argument("--commit", default=None,
                    help="评审目标：指定提交 sha（commit）")
     p.add_argument("--instructions", default=None,
                    help="评审目标：自定义评审指令（custom）")
-    p = sub.add_parser("thread-list"); p.add_argument("--cwd", default=None); p.add_argument("--archived", action="store_true")
+    p = sub.add_parser("thread-list")
-    p = sub.add_parser("thread-read"); p.add_argument("--thread-id", required=True); p.add_argument("--include-turns", action="store_true")
+    p.add_argument("--cwd", default=None)
-    p = sub.add_parser("thread-rename"); p.add_argument("--thread-id", required=True); p.add_argument("--name", required=True)
+    p.add_argument("--archived", action="store_true")
-    p = sub.add_parser("thread-archive"); p.add_argument("--thread-id", required=True)
+    p = sub.add_parser("thread-read")
-    p = sub.add_parser("thread-unarchive"); p.add_argument("--thread-id", required=True)
+    p.add_argument("--thread-id", required=True)
-    p = sub.add_parser("thread-compact"); p.add_argument("--thread-id", required=True)
+    p.add_argument("--include-turns", action="store_true")
-    p = sub.add_parser("skills-list"); p.add_argument("--cwd", default=None)
+    p = sub.add_parser("thread-rename")
+    p.add_argument("--thread-id", required=True)
+    p.add_argument("--name", required=True)
+    p = sub.add_parser("thread-archive")
+    p.add_argument("--thread-id", required=True)
+    p = sub.add_parser("thread-unarchive")
+    p.add_argument("--thread-id", required=True)
+    p = sub.add_parser("thread-compact")
+    p.add_argument("--thread-id", required=True)
+    p = sub.add_parser("skills-list")
+    p.add_argument("--cwd", default=None)
     p = sub.add_parser("apps-list")
     p = sub.add_parser("mcp-servers-list")
     p = sub.add_parser("mcp-servers-reload")
-    p = sub.add_parser("mcp-server-oauth-login"); p.add_argument("--server-name", required=True); p.add_argument("--scopes", default=None); p.add_argument("--timeout-secs", type=int, default=None)
+    p = sub.add_parser("mcp-server-oauth-login")
-    p = sub.add_parser("run"); p.add_argument("--task", required=True); p.add_argument("--prompt", required=True)
+    p.add_argument("--server-name", required=True)
-    p = sub.add_parser("interrupt"); p.add_argument("--task", required=True)
+    p.add_argument("--scopes", default=None)
+    p.add_argument("--timeout-secs", type=int, default=None)
+    p = sub.add_parser("run")
+    p.add_argument("--task", required=True)
+    p.add_argument("--prompt", required=True)
+    p = sub.add_parser("interrupt")
+    p.add_argument("--task", required=True)
     p = sub.add_parser("delegate")
-    p.add_argument("--parent-task", required=True); p.add_argument("--task", required=True)
+    p.add_argument("--parent-task", required=True)
-    p.add_argument("--name", required=True); p.add_argument("--report-to-parent", action="store_true")
+    p.add_argument("--task", required=True)
+    p.add_argument("--name", required=True)
+    p.add_argument("--report-to-parent", action="store_true")
     p = sub.add_parser("goal")
-    p.add_argument("--name", required=True); p.add_argument("--task", required=True)
+    p.add_argument("--name", required=True)
+    p.add_argument("--task", required=True)
     p.add_argument("--repo", default=None)
     p.add_argument("--seed-dir", default=None,
                    help="附件暂存目录：派发时拷进 worktree 再跑 turn")
@@ -2384,9 +2417,13 @@ def main():
                         "轮次与验收；与 --goal-text 同传时忽略（开新目标分支不用它）")
     p.add_argument("--verify-timeout", type=int, default=600,
                    help="验测命令上限（秒）；0 表示不限时，默认 600")
-    p = sub.add_parser("parallel"); p.add_argument("--jobs", required=True)
+    p = sub.add_parser("parallel")
-    p = sub.add_parser("automations"); p.add_argument("--add", action="store_true")
+    p.add_argument("--jobs", required=True)
-    p.add_argument("--name"); p.add_argument("--task"); p.add_argument("--repo")
+    p = sub.add_parser("automations")
+    p.add_argument("--add", action="store_true")
+    p.add_argument("--name")
+    p.add_argument("--task")
+    p.add_argument("--repo")
     p.add_argument("--interval", type=int, default=86400)
     p.add_argument("--cron-line", action="store_true",
                    help="只打印 run-due 的 crontab 行，不装机")
@@ -2395,13 +2432,17 @@ def main():
     p.add_argument("--cron-every-minutes", type=int, default=30)
     sub.add_parser("run-due")
     sub.add_parser("github-workflow")
-    p = sub.add_parser("pr"); p.add_argument("--task", required=True)
+    p = sub.add_parser("pr")
+    p.add_argument("--task", required=True)
     p.add_argument("--repo", default=None)
-    p = sub.add_parser("pr-checks"); p.add_argument("--task", required=True)
+    p = sub.add_parser("pr-checks")
+    p.add_argument("--task", required=True)
     p.add_argument("--repo", default=None)
-    p = sub.add_parser("pr-merge"); p.add_argument("--task", required=True)
+    p = sub.add_parser("pr-merge")
+    p.add_argument("--task", required=True)
     p.add_argument("--repo", default=None)
-    p = sub.add_parser("sleep"); p.add_argument("--seconds", type=float,
+    p = sub.add_parser("sleep")
+    p.add_argument("--seconds", type=float,
                                                 default=5)
 
     args = ap.parse_args()

tests/test_client.py

@@ -0,0 +1,293 @@
+"""client.py 单测（CI 第二阶段覆盖率：基线 0%）。
+
+不跑真实 codex 二进制：AppServer 只跟子进程的 stdin/stdout
+说话，本文件用真实 os.pipe 作 stdout 通道（select + os.read
+原样走）、BytesIO 作 stdin——JSON-RPC 成帧、审批自动回包、
+wait_for 事件分发、initialize/close 都是被测代码的真实逻辑，
+只把“子进程是 codex”换成管道。main() 的分支用一个记录型
+FakeAppServer 测（断言它发出的请求参数与分支选择）。
+"""
+
+import io
+import json
+import os
+import sys
+import tempfile
+import types
+import unittest
+from contextlib import redirect_stdout, redirect_stderr
+from pathlib import Path
+from unittest import mock
+
+REPO_ROOT = Path(__file__).resolve().parents[1]
+sys.path.insert(0, str(REPO_ROOT))
+
+import client as client_mod
+
+
+class FakeProc:
+    """Popen 替身：stdout 是真实 pipe（可 select/os.read），
+    stdin 是 BytesIO（可断言发出的每一行）。"""
+
+    def __init__(self, chunks, wait_exc=None):
+        self.r, self.w = os.pipe()
+        for chunk in chunks:
+            os.write(self.w, chunk)
+        os.close(self.w)
+        self.stdin = io.BytesIO()
+        self.stdout = types.SimpleNamespace(fileno=lambda: self.r)
+        self._wait_exc = wait_exc
+        self.waited = False
+        self.killed = False
+
+    def poll(self):
+        return None
+
+    def wait(self, timeout=None):
+        self.waited = True
+        if self._wait_exc is not None:
+            raise self._wait_exc
+        return 0
+
+    def kill(self):
+        self.killed = True
+
+    def close_fds(self):
+        try:
+            os.close(self.r)
+        except OSError:
+            pass
+
+
+def _line(obj):
+    return (json.dumps(obj) + "\n").encode()
+
+
+class AppServerPipeTest(unittest.TestCase):
+    def _server(self, chunks, codex_home=None, wait_exc=None):
+        proc = FakeProc(chunks, wait_exc=wait_exc)
+        self.addCleanup(proc.close_fds)
+        with mock.patch.object(client_mod.subprocess, "Popen",
+                               return_value=proc):
+            srv = client_mod.AppServer(codex_home=codex_home)
+        srv._proc = proc
+        return srv, proc
+
+    def test_request_and_notify_framing(self):
+        srv, proc = self._server([])
+        rid = srv.request("thread/list", {"a": 1})
+        self.assertEqual(rid, 1)
+        srv.notify("initialized", {})
+        self.assertEqual(srv.next_id, 2)
+        lines = proc.stdin.getvalue().decode().splitlines()
+        self.assertEqual(json.loads(lines[0]),
+                         {"jsonrpc": "2.0", "id": 1, "method": "thread/list",
+                          "params": {"a": 1}})
+        self.assertEqual(json.loads(lines[1]),
+                         {"jsonrpc": "2.0", "method": "initialized",
+                          "params": {}})
+
+    def test_read_msg_answers_approval_then_returns_response(self):
+        chunks = [_line({"jsonrpc": "2.0", "id": 9,
+                         "method": "approval/request",
+                         "params": {"x": 1}}),
+                  _line({"jsonrpc": "2.0", "id": 1, "result": {"ok": True}})]
+        srv, proc = self._server(chunks)
+        srv.approval_policy = "deny"
+        msg = srv._read_msg(5)
+        self.assertEqual(msg, {"jsonrpc": "2.0", "id": 1,
+                               "result": {"ok": True}})
+        sent = proc.stdin.getvalue().decode().splitlines()
+        self.assertEqual(json.loads(sent[0]),
+                         {"jsonrpc": "2.0", "id": 9,
+                          "result": {"decision": "deny"}})
+
+    def test_read_msg_approval_auto(self):
+        chunks = [_line({"jsonrpc": "2.0", "id": 7,
+                         "method": "permission/request", "params": {}}),
+                  _line({"jsonrpc": "2.0", "id": 1, "result": {}})]
+        srv, proc = self._server(chunks)
+        srv.approval_policy = "auto"
+        with redirect_stderr(io.StringIO()):
+            msg = srv._read_msg(5)
+        self.assertEqual(msg["id"], 1)
+        sent = proc.stdin.getvalue().decode().splitlines()
+        self.assertEqual(json.loads(sent[0])["result"],
+                         {"decision": "approve"})
+
+    def test_wait_for_dispatches_events(self):
+        chunks = [_line({"jsonrpc": "2.0", "method": "agent/message/delta",
+                         "params": {"item": {"type": "agentMessage",
+                                             "text": "hi"}}}),
+                  _line({"jsonrpc": "2.0", "id": 3, "result": {"done": 1}})]
+        srv, _ = self._server(chunks)
+        seen = []
+        msg = srv.wait_for(3, timeout=5, on_event=seen.append)
+        self.assertEqual(msg["result"], {"done": 1})
+        self.assertEqual(len(seen), 1)
+        self.assertEqual(seen[0]["method"], "agent/message/delta")
+
+    def test_wait_for_eof_returns_none(self):
+        srv, _ = self._server([])
+        self.assertIsNone(srv.wait_for(1, timeout=5))
+
+    def test_initialize_success_and_error(self):
+        chunks = [_line({"jsonrpc": "2.0", "id": 1,
+                         "result": {"codexHome": "/tmp/ch",
+                                    "platformOs": "linux"}})]
+        srv, proc = self._server(chunks)
+        info = srv.initialize()
+        self.assertEqual(info["platformOs"], "linux")
+        sent = proc.stdin.getvalue().decode().splitlines()
+        self.assertEqual(json.loads(sent[0])["method"], "initialize")
+        self.assertEqual(json.loads(sent[1])["method"], "initialized")
+
+        srv2, _ = self._server(
+            [_line({"jsonrpc": "2.0", "id": 1,
+                    "error": {"code": -1, "message": "boom"}})])
+        with self.assertRaises(RuntimeError):
+            srv2.initialize()
+
+    def test_codex_home_created(self):
+        with tempfile.TemporaryDirectory() as tmp:
+            home = os.path.join(tmp, "new-home")
+            srv, proc = self._server([], codex_home=home)
+            self.assertTrue(os.path.isdir(home))
+            self.assertEqual(proc.stdin.getvalue(), b"")
+
+    def test_close_waits_and_kills_on_timeout(self):
+        srv, proc = self._server([])
+        srv.close()
+        self.assertTrue(proc.waited)
+        self.assertFalse(proc.killed)
+        import subprocess as _sp
+        srv2, proc2 = self._server(
+            [], wait_exc=_sp.TimeoutExpired("codex", 5))
+        srv2.close()
+        self.assertTrue(proc2.killed)
+
+
+class HandleEventTest(unittest.TestCase):
+    def test_agent_delta_printed(self):
+        out = io.StringIO()
+        with redirect_stdout(out):
+            client_mod.handle_event(
+                {"method": "agent/message/delta",
+                 "params": {"item": {"type": "agentMessage",
+                                     "text": "hello"}}})
+        self.assertEqual(out.getvalue(), "hello")
+
+    def test_turn_completed_and_error_paths(self):
+        out, err = io.StringIO(), io.StringIO()
+        with redirect_stdout(out), redirect_stderr(err):
+            client_mod.handle_event(
+                {"method": "turn/completed",
+                 "params": {"status": "completed"}})
+            # 首分支元组外、含 started 的方法走 stderr 分支。
+            client_mod.handle_event(
+                {"method": "session/started", "params": {}})
+            client_mod.handle_event(
+                {"method": "something", "params": {"error": "x"}})
+        self.assertIn("[turn/completed]", out.getvalue())
+        self.assertIn("[session/started]", err.getvalue())
+        self.assertIn("[something]", err.getvalue())
+
+
+class FakeAppServer:
+    def __init__(self, codex_home=None, extra_args=None):
+        self.codex_home = codex_home
+        self.approval_policy = "deny"
+        self.requests = []
+        self.notifies = []
+        self.responses = {}
+        self.read_msgs = []
+        self.closed = False
+
+    def request(self, method, params=None):
+        rid = len(self.requests) + 1
+        self.requests.append((method, params))
+        return rid
+
+    def notify(self, method, params=None):
+        self.notifies.append((method, params))
+
+    def wait_for(self, rid, timeout=60, on_event=None):
+        return self.responses.get(rid)
+
+    def _read_msg(self, timeout):
+        return self.read_msgs.pop(0) if self.read_msgs else None
+
+    def initialize(self):
+        return {"codexHome": "/tmp/ch", "platformOs": "linux"}
+
+    def close(self):
+        self.closed = True
+
+
+class ClientMainTest(unittest.TestCase):
+    def _run(self, argv, setup):
+        fake = FakeAppServer()
+        setup(fake)
+        err = io.StringIO()
+        with mock.patch.object(client_mod, "AppServer",
+                               return_value=fake), \
+                mock.patch.object(sys, "argv", argv), \
+                redirect_stderr(err):
+            out = io.StringIO()
+            with redirect_stdout(out):
+                client_mod.main()
+        return fake, out.getvalue(), err.getvalue()
+
+    def test_list_threads(self):
+        def setup(fake):
+            fake.responses[1] = {"result": {"threads": []}}
+        fake, out, _ = self._run(
+            ["client.py", "--list-threads", "--cwd", "/tmp"],
+            setup)
+        self.assertEqual(fake.requests[0][0], "thread/list")
+        self.assertIn("threads", out)
+        self.assertTrue(fake.closed)
+
+    def test_start_and_turn_completed(self):
+        def setup(fake):
+            fake.responses[1] = {"result": {"thread": {"id": "th-9"},
+                                            "model": "m"}}
+            fake.read_msgs.append(
+                {"method": "turn/completed", "params": {"ok": True}})
+        fake, out, _ = self._run(
+            ["client.py", "--prompt", "fix it", "--cwd", "/tmp",
+             "--model", "gpt-x", "--sandbox", "read-only",
+             "--approval-policy", "never"],
+            setup)
+        start = fake.requests[0]
+        self.assertEqual(start[0], "thread/start")
+        self.assertEqual(start[1]["approvalPolicy"], "never")
+        self.assertEqual(start[1]["sandbox"], "read-only")
+        self.assertEqual(start[1]["model"], "gpt-x")
+        turn = fake.requests[1]
+        self.assertEqual(turn[0], "turn/start")
+        self.assertEqual(turn[1]["threadId"], "th-9")
+        self.assertEqual(turn[1]["input"][0]["text"], "fix it")
+        self.assertIn("[done]", out)
+        self.assertTrue(fake.closed)
+
+    def test_resume_failure_prints_and_returns(self):
+        def setup(fake):
+            fake.responses[1] = {"error": {"message": "gone"}}
+        fake, out, err = self._run(
+            ["client.py", "--thread-id", "th-1", "--cwd", "/tmp"],
+            setup)
+        self.assertEqual(fake.requests[0][0], "thread/resume")
+        self.assertIn("thread/resume", err)
+        self.assertTrue(fake.closed)
+
+    def test_no_prompt_prints_thread_id(self):
+        def setup(fake):
+            fake.responses[1] = {"result": {"thread": {"id": "th-2"}}}
+        _, out, _ = self._run(
+            ["client.py", "--cwd", "/tmp"], setup)
+        self.assertIn("thread_id=th-2", out)
+
+
+if __name__ == "__main__":
+    unittest.main()

tests/test_host.py

@@ -126,7 +126,8 @@ class MergeVerifyTest(unittest.TestCase):
                 "        self.assertEqual(f(), 1)\n")
         (tests_dir / "test_mod.py").write_text(body, encoding="utf-8")
         subprocess.run(["git", "add", "-A"], cwd=tmp, check=True)
-        subprocess.run(["git", "-c", "user.name=Test", "-c", "user.email=test@example.com", "commit", "-qm", "initial"], cwd=tmp,
+        subprocess.run(["git", "-c", "user.name=Test", "-c", "user.email=test@example.com",
+            "commit", "-qm", "initial"], cwd=tmp,
                        check=True)
         # worktree 手动建：模拟 host 的 worktree/分支
         subprocess.run(["git", "worktree", "add", os.path.join(tmp, "wt"),
@@ -175,7 +176,8 @@ class MergeVerifyTest(unittest.TestCase):
             Path(tmp, "mod.py").write_text("def f():\n    return 1\n",
                                            encoding="utf-8")
             subprocess.run(["git", "add", "-A"], cwd=tmp, check=True)
-            subprocess.run(["git", "-c", "user.name=Test", "-c", "user.email=test@example.com", "commit", "-qm", "initial"], cwd=tmp,
+            subprocess.run(["git", "-c", "user.name=Test", "-c", "user.email=test@example.com",
+                "commit", "-qm", "initial"], cwd=tmp,
                            check=True)
             # 分支1：加 add2
             subprocess.run(["git", "worktree", "add",
@@ -643,7 +645,8 @@ class GithubPRTest(unittest.TestCase):
             _init_git_repo(tmp)
             Path(tmp, "f.txt").write_text("x\n")
             subprocess.run(["git", "add", "-A"], cwd=tmp, check=True)
-            subprocess.run(["git", "-c", "user.name=Test", "-c", "user.email=test@example.com", "commit", "-qm", "initial"],
+            subprocess.run(["git", "-c", "user.name=Test", "-c", "user.email=test@example.com",
+                "commit", "-qm", "initial"],
                            cwd=tmp, check=True)
             task = {"name": "t", "branch": "host/t", "status": "completed",
                     "prompt": "do it"}
@@ -658,12 +661,14 @@ class GithubPRTest(unittest.TestCase):
             _init_git_repo(repo)
             Path(repo, "f.txt").write_text("x\n")
             subprocess.run(["git", "add", "-A"], cwd=repo, check=True)
-            subprocess.run(["git", "-c", "user.name=Test", "-c", "user.email=test@example.com", "commit", "-qm", "initial"],
+            subprocess.run(["git", "-c", "user.name=Test", "-c", "user.email=test@example.com",
+                "commit", "-qm", "initial"],
                            cwd=repo, check=True)
             subprocess.run(["git", "checkout", "-qb", "host/t1"],
                            cwd=repo, check=True)
             Path(repo, "f.txt").write_text("y\n")
-            subprocess.run(["git", "-c", "user.name=Test", "-c", "user.email=test@example.com", "commit", "-qam", "change"],
+            subprocess.run(["git", "-c", "user.name=Test", "-c", "user.email=test@example.com",
+                "commit", "-qam", "change"],
                            cwd=repo, check=True)
             remote = os.path.join(tmp, "remote.git")
             subprocess.run(["git", "init", "--bare", "-q", remote],

tests/test_host_dispatch_branch.py

@@ -0,0 +1,94 @@
+"""dispatch_from_branch 回归测试（CI 第二阶段 F841 证据）。
+
+背景：host.py 的 dispatch_from_branch 里 `git commit` 的返回值
+曾以 `c = git(...)` 绑定却从未读取（F841）。本轨只去绑定、
+不删调用——提交本身是依赖派发的前置：依赖 worktree 的改动
+必须先落到依赖分支，新 worktree 才从该分支建出、带血缘。
+本测试用真实 git 仓库断言这条链：
+
+- 依赖 worktree 里未提交的改动被提交到依赖分支；
+- 新 worktree 的分支 host/<name> 由依赖分支 HEAD 派生
+  （含该提交），且任务记录进 state。
+"""
+
+import os
+import shutil
+import subprocess
+import sys
+import tempfile
+import unittest
+import uuid
+from pathlib import Path
+
+REPO_ROOT = Path(__file__).resolve().parents[1]
+sys.path.insert(0, str(REPO_ROOT))
+
+import host
+
+
+def _git(args, cwd, check=True):
+    return subprocess.run(["git"] + args, cwd=cwd, check=check,
+                          capture_output=True, text=True)
+
+
+class FakeServer:
+    def create_thread(self, path, model=None, sandbox=None,
+                      approval_policy=None):
+        self.created = (path, model, sandbox, approval_policy)
+        return "th-1"
+
+
+class DispatchFromBranchTest(unittest.TestCase):
+    def test_commit_then_worktree_from_dep_branch(self):
+        name = "dep-" + uuid.uuid4().hex[:8]
+        server = FakeServer()
+        state = {"tasks": {}}
+        with tempfile.TemporaryDirectory() as tmp:
+            repo = os.path.join(tmp, "repo")
+            os.makedirs(repo)
+            _git(["init", "-q", "-b", "main"], repo)
+            _git(["config", "user.name", "Test"], repo)
+            _git(["config", "user.email", "test@example.com"], repo)
+            Path(repo, "f.txt").write_text("base\n", encoding="utf-8")
+            _git(["add", "-A"], repo)
+            _git(["commit", "-qm", "initial"], repo)
+            _git(["branch", "dep-branch"], repo)
+            dep_wt = os.path.join(tmp, "dep-wt")
+            _git(["worktree", "add", "-q", dep_wt, "dep-branch"], repo)
+            Path(dep_wt, "f.txt").write_text("dep change\n",
+                                             encoding="utf-8")
+            try:
+                task, err = host.dispatch_from_branch(
+                    server, state, name, "do it", repo, "dep-branch",
+                    dep_wt, "model-x", "medium", "workspace-write", "never")
+                self.assertEqual(err, "")
+                self.assertIsNotNone(task)
+                self.assertEqual(task["branch"], "host/" + name)
+                self.assertEqual(task["thread_id"], "th-1")
+                # 依赖分支 HEAD 含本次提交，且新分支由它派生。
+                dep_head = _git(["rev-parse", "dep-branch"],
+                                repo).stdout.strip()
+                new_head = _git(["rev-parse", "host/" + name],
+                                repo).stdout.strip()
+                self.assertEqual(new_head, dep_head)
+                msg = _git(["log", "-1", "--format=%s", "dep-branch"],
+                           repo).stdout.strip()
+                self.assertIn(name, msg)
+                self.assertEqual(
+                    Path(task["cwd"], "f.txt").read_text(encoding="utf-8"),
+                    "dep change\n")
+                self.assertIs(state["tasks"][name], task)
+            finally:
+                # 清理 HERE/work 下建出的 worktree（work/ 已
+                # gitignore，但不留给下一轮测试；须在临时仓
+                # 删除前用该仓的 git 登记移除）。
+                path = os.path.join(host.HERE, "work",
+                                    "host-worktrees", name)
+                if os.path.exists(path):
+                    _git(["worktree", "remove", "--force", path], repo,
+                         check=False)
+                    shutil.rmtree(path, ignore_errors=True)
+
+
+if __name__ == "__main__":
+    unittest.main()

tests/test_host_run_verify.py

@@ -0,0 +1,53 @@
+"""host.run_verify / goal verify 的 shell 语义证据（CI 第二阶段
+bandit High triage，第 3 项白名单轨范围外、本轨只做 triage）。
+
+背景：bandit High 4 项全是 host.py 的 shell=True：
+run_verify（1319）、notify（1497）、goal verify 两处
+（2099/2103）。本轨不改它们，理由须有测试证据：
+
+- run_verify 的命令是整串 shell 字符串（配置/默认验测命令），
+  管道与复合命令必须经 shell 解释：列表式 exec 形跑不通。
+  本文件断言管道与 && 真实生效——改 shell=False 会立刻失败。
+- notify 的插值已全部 shlex.quote：证据在 tests/test_host.py
+  的 test_notify_failure_quotes_shell_metacharacters（`; touch`
+  与 `$(touch )` 不执行、只落日志原文），本文件不再重复。
+"""
+
+import sys
+import tempfile
+import unittest
+from pathlib import Path
+
+REPO_ROOT = Path(__file__).resolve().parents[1]
+sys.path.insert(0, str(REPO_ROOT))
+
+import host
+
+
+class RunVerifyShellTest(unittest.TestCase):
+    def test_pipe_requires_shell(self):
+        with tempfile.TemporaryDirectory() as tmp:
+            rc, out = host.run_verify("echo hello | tr a-z A-Z", tmp)
+            self.assertEqual(rc, 0)
+            self.assertEqual(out, "HELLO")
+
+    def test_compound_command_requires_shell(self):
+        with tempfile.TemporaryDirectory() as tmp:
+            rc, out = host.run_verify("echo a && echo b", tmp)
+            self.assertEqual(rc, 0)
+            self.assertEqual(out, "a\nb")
+
+    def test_empty_command_is_noop(self):
+        with tempfile.TemporaryDirectory() as tmp:
+            self.assertEqual(host.run_verify("", tmp), (0, ""))
+            self.assertEqual(host.run_verify(None, tmp), (0, ""))
+
+    def test_failing_command_returns_rc_and_output(self):
+        with tempfile.TemporaryDirectory() as tmp:
+            rc, out = host.run_verify("echo oops; exit 3", tmp)
+            self.assertEqual(rc, 3)
+            self.assertEqual(out, "oops")
+
+
+if __name__ == "__main__":
+    unittest.main()

tests/test_host_trace.py

@@ -13,11 +13,9 @@ import unittest
 from pathlib import Path
 
 try:
-    import host
     from host import Server, TRACE_MAX_LINES, _summarize_trace_item
 except ImportError:
     sys.path.insert(0, str(Path(__file__).resolve().parents[1]))
-    import host
     from host import Server, TRACE_MAX_LINES, _summarize_trace_item
 
 

tests/test_run_parallel.py

@@ -0,0 +1,105 @@
+"""run_parallel.py 单测（CI 第二阶段覆盖率：基线 0%）。
+
+不跑真实 client.py/codex：只断言 main() 拼出的子进程命令
+（可执行文件、client.py 路径、cwd 解析、sandbox 默认值、
+approval-policy、codex-home 透传）与成功计数/返回码——
+命令构造与汇总正是本文件的全部逻辑。
+"""
+
+import io
+import json
+import os
+import sys
+import tempfile
+import unittest
+from contextlib import redirect_stdout
+from pathlib import Path
+from unittest import mock
+
+REPO_ROOT = Path(__file__).resolve().parents[1]
+sys.path.insert(0, str(REPO_ROOT))
+
+import run_parallel as rp_mod
+
+
+class FakePopen:
+    seen = []
+
+    def __init__(self, cmd):
+        self.cmd = cmd
+        FakePopen.seen.append(cmd)
+        self._rc = 0
+
+    def wait(self):
+        return self._rc
+
+
+class RunParallelTest(unittest.TestCase):
+    def _run(self, jobs, extra_args):
+        FakePopen.seen = []
+        with tempfile.TemporaryDirectory() as tmp:
+            jobs_path = os.path.join(tmp, "jobs.json")
+            with open(jobs_path, "w", encoding="utf-8") as f:
+                json.dump(jobs, f)
+            argv = ["run_parallel.py", "--jobs", jobs_path] + extra_args
+            with mock.patch.object(rp_mod.subprocess, "Popen", FakePopen), \
+                    mock.patch.object(sys, "argv", argv):
+                out = io.StringIO()
+                with redirect_stdout(out):
+                    rc = rp_mod.main()
+        return rc, list(FakePopen.seen), out.getvalue()
+
+    def test_commands_and_all_success(self):
+        jobs = [{"cwd": "job-a", "prompt": "review a"},
+                {"cwd": "/tmp/abs-b", "prompt": "review b",
+                 "sandbox": "workspace-write"}]
+        rc, seen, out = self._run(jobs, ["--codex-home", "/tmp/ch"])
+        self.assertEqual(rc, 0)
+        self.assertEqual(len(seen), 2)
+        client = os.path.join(rp_mod.HERE, "client.py")
+        self.assertEqual(seen[0], [sys.executable, client,
+                                   "--prompt", "review a",
+                                   "--cwd", os.path.join(rp_mod.HERE, "job-a"),
+                                   "--sandbox", "read-only",
+                                   "--approval-policy", "never",
+                                   "--codex-home", "/tmp/ch"])
+        self.assertEqual(seen[1][5], "/tmp/abs-b")
+        self.assertEqual(seen[1][7], "workspace-write")
+        self.assertIn("2/2", out)
+
+    def test_one_failure_returns_1(self):
+        class FailingPopen(FakePopen):
+            def __init__(self, cmd):
+                super().__init__(cmd)
+                self._idx = len(FakePopen.seen)
+
+            def wait(self):
+                return 1 if self._idx == 2 else 0
+
+        jobs = [{"cwd": "/tmp/a", "prompt": "x"},
+                {"cwd": "/tmp/b", "prompt": "y"}]
+        FakePopen.seen = []
+        with tempfile.TemporaryDirectory() as tmp:
+            jobs_path = os.path.join(tmp, "jobs.json")
+            with open(jobs_path, "w", encoding="utf-8") as f:
+                json.dump(jobs, f)
+            with mock.patch.object(rp_mod.subprocess, "Popen",
+                                   FailingPopen), \
+                    mock.patch.object(sys, "argv",
+                                      ["run_parallel.py", "--jobs",
+                                       jobs_path]):
+                out = io.StringIO()
+                with redirect_stdout(out):
+                    rc = rp_mod.main()
+        self.assertEqual(rc, 1)
+        self.assertIn("1/2", out.getvalue())
+
+    def test_empty_jobs_returns_0(self):
+        rc, seen, out = self._run([], [])
+        self.assertEqual(rc, 0)
+        self.assertEqual(seen, [])
+        self.assertIn("0/0", out)
+
+
+if __name__ == "__main__":
+    unittest.main()