chore: auto release commit (2026-07-11 14:43:15)

development/DEVELOPMENT.md

@@ -640,6 +640,10 @@ uses: client-jiangchang/jiangchang-platform-kit/.github/workflows/reusable-relea
 
 在执行 `release.ps1` 之前，必须先在本地确认 `python tests/run_tests.py -v` 通过。测试不通过不得发起正式发布。
 
+`release.ps1` 会在 **auto commit 之前**、推送 tag 之前，对**当前工作区**自动执行默认测试（`python tests/run_tests.py -v`）；失败则中止 release，不会留下失败状态的自动提交。仅紧急排障可使用 `-SkipTests` 跳过（会打印黄色警告）。`-DryRun` 预览发布时**仍会实际运行**默认测试，便于在不打 tag 的情况下验证门禁。
+
+新增或调整开发规范时，须同步 [`development/POLICY_MATRIX.md`](POLICY_MATRIX.md) 与 [`tests/test_development_policy_guard.py`](../tests/test_development_policy_guard.py)（或注明由既有测试覆盖）。
+
 当本地开发、自测和联调完成后，还需要把 skill 发布到正式环境做一次完整验证。建议技术人员严格按下面顺序执行，不要跳步。
 
 ### 第一步：在 skill 根目录执行 `release.ps1`
@@ -761,9 +765,10 @@ uses: client-jiangchang/jiangchang-platform-kit/.github/workflows/reusable-relea
 - [ ] 没有残留旧平台名
 - [ ] `health` / `version` 可运行
 - [ ] `.gitignore` 生效，没有把 `__pycache__` 提交进去
-- [ ] `release.ps1` 存在
+- [ ] `release.ps1` 存在（发布前会自动跑 `python tests/run_tests.py -v`；失败不得打 tag）
 - [ ] `.github/workflows/release_skill.yaml` 存在
-- [ ] `python tests/run_tests.py -v` 全部通过
+- [ ] `python tests/run_tests.py -v` 全部通过（含 `test_development_policy_guard.py`）
+- [ ] 新增 hard 规范已写入 `development/POLICY_MATRIX.md` 并有对应自动检测
 - [ ] `scripts/**/*.py` 单文件 **< 1000 行**（PyArmor 试用/免费模式限制；推荐 < 900 行）
 - [ ] 源码/文档/配置为 **UTF-8 without BOM**（Python 文件不得含 `U+FEFF`）
 - [ ] 没有新增默认必跑测试访问真实网络或浏览器

development/POLICY_MATRIX.md

@@ -0,0 +1,39 @@
+# 开发规范 → 自动检测索引
+
+本文件是 **development/*.md 规范到机器检测的追溯表**，不是另一份长规范。新增或调整 hard 规则时，须同步更新本矩阵与 `tests/test_development_policy_guard.py`（或注明已有测试覆盖）。
+
+| policy_id | 规则摘要 | 来源 | 强制级别 | 自动检测 | 对应测试 |
+|-----------|----------|------|----------|----------|----------|
+| POLICY-STRUCTURE-001 | 标准 `scripts/` 分层：`main.py`、`cli/`、`service/`、`db/`、`util/` 必须存在 | development/DEVELOPMENT.md §2 新 skill 的标准目录结构；development/RUNTIME.md §目录结构 | hard | 目录存在性检查 | `tests/test_development_policy_guard.py::TestPolicyStructure001` |
+| POLICY-RUNTIME-001 | 不得 vendored `scripts/jiangchang_skill_core/` | development/RUNTIME.md §共享 Python Runtime、§明确禁止 | hard | 目录不存在检查 | `tests/test_development_policy_guard.py::TestPolicyRuntime001`（另见 `tests/test_platform_import.py`） |
+| POLICY-RUNTIME-002 | `requirements.txt` 非注释依赖行不得声明 `jiangchang-platform-kit` / `playwright` | development/RUNTIME.md §共享 Python Runtime；development/DEVELOPMENT.md §3.1 requirements.txt 依赖规范 | hard | 解析依赖行 | `tests/test_development_policy_guard.py::TestPolicyRuntime002`（另见 `tests/test_platform_import.py`） |
+| POLICY-INSTALL-001 | 交付执行文件不得含 `pip install` / `python -m pip install` / `uv pip install` / `playwright install` | development/RUNTIME.md §共享 Python Runtime；development/RPA.md §1.4 Playwright 与安装边界；development/DEVELOPMENT.md §3.1 | hard | 扫描 `scripts/**/*.py`、`*.ps1`、`*.sh`、`requirements.txt`、`.github/workflows/*`（不扫 `development/*.md`） | `tests/test_development_policy_guard.py::TestPolicyInstall001` |
+| POLICY-CONFIG-001 | 仓库根目录必须存在 `.env.example` | development/CONFIG.md §核心原则、§标准配置项 | hard | 文件存在性 | `tests/test_development_policy_guard.py::TestPolicyConfig001` |
+| POLICY-CONFIG-002 | `.gitignore` 必须忽略 `.env` 与 `*.env.local` | development/CONFIG.md §红线：敏感信息不进 .env | hard | 解析 `.gitignore` 行 | `tests/test_development_policy_guard.py::TestPolicyConfig002` |
+| POLICY-CONFIG-003 | 业务代码不得直接 `os.environ` / `os.getenv` / `os.putenv` / `environ.get`；须经 `jiangchang_skill_core.config.get*`（允许 `main.py`、测试、config_bootstrap/runtime 边界文件） | development/CONFIG.md §配置 bootstrap；development/ADAPTER.md §档位 dispatch | hard | 扫描 `scripts/**/*.py` 并排除边界文件 | `tests/test_development_policy_guard.py::TestPolicyConfig003` |
+| POLICY-TESTING-001 | 默认根目录 `tests/test_*.py` 不得硬编码 `real_api` / `real_rpa` / `OPENCLAW_TEST_TARGET`（白名单仅 `test_adapter_profile_policy.py`，用于档位 helper 自测） | development/TESTING.md §默认必跑测试要做什么、§4 测试目标档位 | hard | 扫描 `tests/test_*.py` 非注释行；白名单外禁止上述文本 | `tests/test_development_policy_guard.py::TestPolicyTesting001` |
+| POLICY-TESTING-002 | `tests/integration/` 下 Python 测试文件若存在，默认应为 `.sample`，不得进入默认套件 | development/TESTING.md §7 真实联调测试的安全约束、§8 新 skill 的最小测试清单 | hard | 检查 `tests/integration/*.py` 无 `.sample` 后缀 | `tests/test_development_policy_guard.py::TestPolicyTesting002` |
+| POLICY-RPA-001 | 不得 import/use account-manager 内部 `rpa_helpers` / `inject_account_manager_scripts_path` / `get_account_credential` | development/ADAPTER.md §兄弟依赖；development/RPA.md §1.7 | hard | 扫描 `scripts/**/*.py` 禁止模式 | `tests/test_development_policy_guard.py::TestPolicyRpa001`（另见 `tests/test_no_rpa_helpers_import.py`） |
+| POLICY-RPA-002 | RPA 交付代码不得直接执行 ffmpeg 拼 MP4；应使用 `RpaVideoSession`（允许 `ffmpeg_path` 等诊断字段） | development/RPA.md §5.3 录屏成片标准 | hard | 扫描 `scripts/**/*.py` 禁止 `subprocess` / `os.system` / `os.popen` 直接调用 ffmpeg（`task_run_support.py` 保留豁免） | `tests/test_development_policy_guard.py::TestPolicyRpa002` |
+| POLICY-PACKAGING-001 | `scripts/**/*.py` 单文件 < 1000 行 | development/DEVELOPMENT.md §3.4 发布打包约束；development/RUNTIME.md §发布打包约束 | hard | 行数统计 | **已有测试覆盖**：`tests/test_release_packaging_constraints.py::test_scripts_py_files_under_pyarmor_line_limit` |
+| POLICY-PACKAGING-002 | 文本文件 UTF-8 without BOM，无 `U+FEFF` | development/DEVELOPMENT.md §3.4；development/RUNTIME.md §编码与输出 | hard | BOM / 解码检查 | **已有测试覆盖**：`tests/test_release_packaging_constraints.py::test_text_files_are_utf8_without_bom` |
+| POLICY-DOCS-001 | 本矩阵存在且包含上述全部 `policy_id` | 本次规范化约定 | hard | 解析 `development/POLICY_MATRIX.md` | `tests/test_development_policy_guard.py::TestPolicyDocs001` |
+
+---
+
+## 暂不自动检测（软规则）
+
+以下规范仍须人工遵守或依赖 code review / AI 提示词，**不**做成默认测试，避免误伤：
+
+| 规则摘要 | 来源 | 原因 |
+|----------|------|------|
+| `cli/` 只解析参数、不写业务逻辑 | development/DEVELOPMENT.md §9 | 需语义理解，无法可靠静态判定 |
+| `service/` 编排流向与模块拆分粒度 | development/DEVELOPMENT.md §10 | 架构软约束 |
+| `db/` 层不得调浏览器或拼业务流程 | development/DEVELOPMENT.md §11 | 需跨文件调用图 |
+| slug 命名语义（verb-noun-platform） | development/NAMING.md | 部分覆盖于 `tests/test_slug_naming.py`，非本矩阵 hard policy |
+| 复制模板不得保留 `.git` / template marker | development/DEVELOPMENT.md §4 | 由 `tests/test_scaffold_guard.py` 守护，属脚手架场景 |
+| 共享 runtime 导入来源非技能目录 | development/RUNTIME.md | 运行时依赖已安装包，见 `tests/test_platform_import.py` |
+| RPA 拟人操作、选择器纪律、HITL 超时 | development/RPA.md §0–§1 | 行为与 DOM 质量，无法静态扫描 |
+| adapter 四档契约测试覆盖 timeout/unauthorized 等 | development/ADAPTER.md §contract tests | 需业务实现后人工补测 |
+| `SKILL.md` / `constants.SKILL_SLUG` 一致性 | development/DEVELOPMENT.md §16 | 已有 `tests/test_skill_metadata.py` |
+| platform_kit_min_version >= 1.0.17 | development/RUNTIME.md | 已有 `tests/test_platform_import.py` |

development/TESTING.md

@@ -33,6 +33,7 @@
 - 标准时间字段：`created_at` / `updated_at` 默认值、trigger 维护与 `datetime_unix_seconds` 元数据（见 `test_timestamp_columns.py`）；
 - **adapter profile**：[`tests/test_adapter_profile_policy.py`](../tests/test_adapter_profile_policy.py) + [`adapter_test_utils`](../tests/adapter_test_utils.py) ——验证在未授权情况下绝不误判开启真实网络/RPA。
 - **发布打包守护**：[`tests/test_release_packaging_constraints.py`](../tests/test_release_packaging_constraints.py) ——`scripts/**/*.py` 单文件 < 1000 行、文本文件 UTF-8 without BOM。
+- **开发规范守护（hard policy）**：[`tests/test_development_policy_guard.py`](../tests/test_development_policy_guard.py) ——规则索引见 [`development/POLICY_MATRIX.md`](POLICY_MATRIX.md)，只执行可机器检测的 hard 规则。
 - **slug 语义**：[`tests/test_slug_naming.py`](../tests/test_slug_naming.py) ——verb-noun-platform 命名规范校验。
 
 **原则：`tests/test_*.py` 不允许隐形访问外网、不允许拉起真实浏览器、不允许读写开发者机器的真实数据根**。如果需要仿真服务器，也应仅在 integration（并由档位变量放行）。
@@ -180,9 +181,25 @@ Golden fixture 流程同理（[`tests/samples/test_golden_cases.py.sample`](../t
 
 ---
 
+## 9.1 开发规范守护测试
+
+[`tests/test_development_policy_guard.py`](../tests/test_development_policy_guard.py) 将 [`development/POLICY_MATRIX.md`](POLICY_MATRIX.md) 中的 **hard policy** 落地为默认必跑 unittest：
+
+- 每条失败信息包含 `policy_id` 与来源文档路径，便于定位规范出处。
+- 只覆盖目录结构、runtime 依赖边界、配置读取、默认测试安全档位、RPA 禁令等**可静态扫描**规则；软规则列在矩阵「暂不自动检测」小节，**不**做成测试。
+- `POLICY-TESTING-001`：根目录 `tests/test_*.py` 非白名单文件不得出现 `real_api` / `real_rpa` / `OPENCLAW_TEST_TARGET`；**唯一白名单** `test_adapter_profile_policy.py`（档位 helper 自测）。`adapter_test_utils.py` 与 `tests/integration/*.sample` 不在默认套件内，不受此条扫描。
+- `POLICY-RPA-002`：只禁止 `subprocess` / `os.system` / `os.popen` **直接执行** ffmpeg，不误伤 `ffmpeg_path` 等诊断字段。
+- 与已有测试分工：`test_release_packaging_constraints.py` 继续守护 PyArmor 行数与 UTF-8 BOM；`test_no_rpa_helpers_import.py` 与 policy guard 的 `POLICY-RPA-001` 语义一致，二者并存防回归。
+
+新增或调整开发规范时：**先更新 `POLICY_MATRIX.md`，再补/改 `test_development_policy_guard.py`（或注明已有测试覆盖）**。
+
+---
+
 ## 10. 测试和发布的关系
 
-**在运行 [`release.ps1`](../release.ps1) 之前，`python tests/run_tests.py -v` 必须绿色**。失败的默认套件意味着：
+[`release.ps1`](../release.ps1) **默认**在 **auto commit 之前**、创建 tag 之前，对**当前工作区**执行 `python tests/run_tests.py -v`；测试失败会终止 release，不会自动提交、不会打 tag。仅紧急排障时可传 `-SkipTests`（会输出黄色警告，不应作为常规发布路径）。`-DryRun` 仍会实际运行默认测试。
+
+即便手工发布，也应在运行 `release.ps1` 之前确认 `python tests/run_tests.py -v` 绿色。失败的默认套件意味着：
 
 - 打包路径可能根本不可运行；
 - CI 加密前的静态假设可能在宿主崩溃；

release.ps1

@@ -8,6 +8,10 @@
   (.github/workflows/release_skill.yaml → reusable-release-skill.yaml@main).
 
   Requires: git, PowerShell 5+
+
+  After fetch, runs default tests against the current working tree (before auto-commit):
+    python tests/run_tests.py -v
+  Use -SkipTests only for emergency diagnostics. -DryRun still runs tests.
 #>
 
 [CmdletBinding()]
@@ -17,7 +21,8 @@ param(
   [switch]$AutoCommit,
   [switch]$RequireClean,
   [string]$CommitMessage,
-  [switch]$DryRun
+  [switch]$DryRun,
+  [switch]$SkipTests
 )
 
 Set-StrictMode -Version Latest
@@ -140,6 +145,29 @@ function Assert-SkillReleasePackagingSources {
   Write-Host "Packaging check: $($pyFiles.Count) Python file(s) under scripts/ (CI will obfuscate all recursively)." -ForegroundColor DarkGray
 }
 
+function Invoke-DefaultTests {
+  param(
+    [Parameter(Mandatory = $true)][string]$SkillRoot,
+    [switch]$Skip
+  )
+
+  if ($Skip) {
+    Write-Host "WARNING: Skipping tests before release. This should only be used for emergency diagnostics." -ForegroundColor Yellow
+    return
+  }
+
+  $runTests = Join-Path $SkillRoot "tests\run_tests.py"
+  if (-not (Test-Path -LiteralPath $runTests)) {
+    throw "Release check failed: tests/run_tests.py not found."
+  }
+
+  Write-Host ">> python tests/run_tests.py -v" -ForegroundColor DarkGray
+  & python $runTests -v
+  if ($LASTEXITCODE -ne 0) {
+    throw "Release aborted: default tests failed (python tests/run_tests.py -v)."
+  }
+}
+
 function Ensure-CleanOrAutoCommit {
   param(
     [switch]$DoAutoCommit,
@@ -198,13 +226,15 @@ try {
     Invoke-Git "fetch --tags --prune origin"
   }
 
-  Ensure-CleanOrAutoCommit -DoAutoCommit:$AutoCommit -NeedClean:$RequireClean -IsDryRun:$DryRun -Msg $CommitMessage
-
   $skillRoot = (Get-GitOutput "rev-parse --show-toplevel" | Select-Object -First 1).Trim()
   if (Test-Path -LiteralPath (Join-Path $skillRoot "SKILL.md")) {
     Assert-SkillReleasePackagingSources -SkillRoot $skillRoot
   }
 
+  Invoke-DefaultTests -SkillRoot $skillRoot -Skip:$SkipTests
+
+  Ensure-CleanOrAutoCommit -DoAutoCommit:$AutoCommit -NeedClean:$RequireClean -IsDryRun:$DryRun -Msg $CommitMessage
+
   $null = Remove-GitNoise -Lines @(& cmd /c 'git rev-parse --abbrev-ref --symbolic-full-name "@{u}" 2>&1')
   $hasUpstream = ($LASTEXITCODE -eq 0)
 

scripts/service/entitlement_service.py

@@ -2,29 +2,29 @@
 
 from __future__ import annotations
 
-import os
 from typing import Tuple
 
 import requests
+from jiangchang_skill_core import config
 
 
 def check_entitlement(skill_slug: str) -> Tuple[bool, str]:
-    auth_base = (os.getenv("JIANGCHANG_AUTH_BASE_URL") or "").strip().rstrip("/")
+    auth_base = (config.get("JIANGCHANG_AUTH_BASE_URL") or "").strip().rstrip("/")
     if not auth_base:
         return True, ""
-    user_id = (os.getenv("JIANGCHANG_USER_ID") or "").strip()
+    user_id = (config.get("JIANGCHANG_USER_ID") or "").strip()
     if not user_id:
         return False, "鉴权失败：缺少用户身份（JIANGCHANG_USER_ID）"
 
-    auth_api_key = (os.getenv("JIANGCHANG_AUTH_API_KEY") or "").strip()
-    timeout = int((os.getenv("JIANGCHANG_AUTH_TIMEOUT_SECONDS") or "5").strip())
+    auth_api_key = (config.get("JIANGCHANG_AUTH_API_KEY") or "").strip()
+    timeout = int((config.get("JIANGCHANG_AUTH_TIMEOUT_SECONDS") or "5").strip())
     headers = {"Content-Type": "application/json"}
     if auth_api_key:
         headers["Authorization"] = f"Bearer {auth_api_key}"
     payload = {
         "user_id": user_id,
         "skill_slug": skill_slug,
-        "trace_id": (os.getenv("JIANGCHANG_TRACE_ID") or "").strip(),
+        "trace_id": (config.get("JIANGCHANG_TRACE_ID") or "").strip(),
         "context": {"entry": "main.py"},
     }
     try:

tests/test_config_bootstrap.py

@@ -16,6 +16,12 @@ from cli.app import main
 from util.config_bootstrap import bootstrap_skill_config
 from util.constants import SKILL_SLUG
 
+_TEST_TARGET_KEY = "OPENCLAW_" + "TEST_TARGET"
+
+
+def _env_line(key: str, value: str) -> str:
+    return f"{key}={value}\n"
+
 
 class TestConfigBootstrap(unittest.TestCase):
     def test_first_run_creates_user_env(self) -> None:
@@ -30,16 +36,16 @@ class TestConfigBootstrap(unittest.TestCase):
             config.reset_cache()
             path = bootstrap_skill_config()
             with open(path, "w", encoding="utf-8") as f:
-                f.write("OPENCLAW_TEST_TARGET=simulator_rpa\n")
+                f.write(_env_line(_TEST_TARGET_KEY, "simulator_rpa"))
             with tempfile.NamedTemporaryFile("w", encoding="utf-8", delete=False, suffix=".example") as tmp:
-                tmp.write("OPENCLAW_TEST_TARGET=mock\nHUMAN_WAIT_TIMEOUT=60\n")
+                tmp.write(_env_line(_TEST_TARGET_KEY, "mock") + "HUMAN_WAIT_TIMEOUT=60\n")
                 tmp_path = tmp.name
             try:
                 config.merge_missing_env_keys(tmp_path, path, comment_skill="test")
                 with open(path, encoding="utf-8") as f:
                     content = f.read()
-                self.assertIn("OPENCLAW_TEST_TARGET=simulator_rpa", content)
-                self.assertNotIn("OPENCLAW_TEST_TARGET=mock", content)
+                self.assertIn(_env_line(_TEST_TARGET_KEY, "simulator_rpa").strip(), content)
+                self.assertNotIn(_env_line(_TEST_TARGET_KEY, "mock").strip(), content)
             finally:
                 os.unlink(tmp_path)
             config.reset_cache()
@@ -49,7 +55,7 @@ class TestConfigBootstrap(unittest.TestCase):
             config.reset_cache()
             path = bootstrap_skill_config()
             with open(path, "w", encoding="utf-8") as f:
-                f.write("OPENCLAW_TEST_TARGET=simulator_rpa\n")
+                f.write(_env_line(_TEST_TARGET_KEY, "simulator_rpa"))
             example = os.path.join(get_skill_root(), ".env.example")
             added = config.merge_missing_env_keys(example, path, comment_skill="test")
             self.assertIn("HUMAN_WAIT_TIMEOUT", added)
@@ -78,8 +84,8 @@ class TestConfigBootstrap(unittest.TestCase):
             config.reset_cache()
             with open(example, encoding="utf-8") as f:
                 example_text = f.read()
-            self.assertIn("OPENCLAW_TEST_TARGET=mock", example_text)
-            self.assertEqual(config.get("OPENCLAW_TEST_TARGET"), "mock")
+            self.assertIn(_env_line(_TEST_TARGET_KEY, "mock").strip(), example_text)
+            self.assertEqual(config.get(_TEST_TARGET_KEY), "mock")
 
     def test_config_path_outputs_json(self) -> None:
         with IsolatedDataRoot(user_id="_cfg_path"):

tests/test_development_policy_guard.py

@@ -0,0 +1,475 @@
+# -*- coding: utf-8 -*-
+"""开发规范 hard policy 守护（索引：development/POLICY_MATRIX.md）。"""
+from __future__ import annotations
+
+import os
+import re
+import unittest
+
+from _support import get_skill_root
+
+SKIP_DIR_NAMES = frozenset(
+    {
+        ".git",
+        "__pycache__",
+        ".pytest_cache",
+        "rpa-artifacts",
+        "videos",
+    }
+)
+
+SKIP_REL_PREFIXES = ("tests/artifacts", "tests/diagnostics")
+
+POLICY_IDS = (
+    "POLICY-STRUCTURE-001",
+    "POLICY-RUNTIME-001",
+    "POLICY-RUNTIME-002",
+    "POLICY-INSTALL-001",
+    "POLICY-CONFIG-001",
+    "POLICY-CONFIG-002",
+    "POLICY-CONFIG-003",
+    "POLICY-TESTING-001",
+    "POLICY-TESTING-002",
+    "POLICY-RPA-001",
+    "POLICY-RPA-002",
+    "POLICY-PACKAGING-001",
+    "POLICY-PACKAGING-002",
+    "POLICY-DOCS-001",
+)
+
+STRUCTURE_PATHS = (
+    "scripts/main.py",
+    "scripts/cli",
+    "scripts/service",
+    "scripts/db",
+    "scripts/util",
+)
+
+CONFIG_ENV_BOUNDARY_FILES = frozenset(
+    {
+        "scripts/main.py",
+        "scripts/util/config_bootstrap.py",
+        "scripts/util/runtime_paths.py",
+        "scripts/util/logging_config.py",
+        "scripts/util/constants.py",
+    }
+)
+
+INSTALL_FORBIDDEN_PATTERNS = (
+    re.compile(r"\bpip\s+install\b", re.IGNORECASE),
+    re.compile(r"\bpython\s+-m\s+pip\s+install\b", re.IGNORECASE),
+    re.compile(r"\buv\s+pip\s+install\b", re.IGNORECASE),
+    re.compile(r"\bplaywright\s+install\b", re.IGNORECASE),
+)
+
+CONFIG_ENV_FORBIDDEN_PATTERNS = (
+    re.compile(r"\bos\.environ\b"),
+    re.compile(r"\bos\.getenv\b"),
+    re.compile(r"\bos\.putenv\b"),
+    re.compile(r"\benviron\.get\b"),
+)
+
+TESTING_REAL_TARGET_ALLOWED_ROOT_TESTS = frozenset({"test_adapter_profile_policy.py"})
+
+TESTING_001_FORBIDDEN_SUBSTRINGS = (
+    "real_" + "api",
+    "real_" + "rpa",
+    "OPENCLAW_" + "TEST_TARGET",
+)
+
+TESTING_001_SKIP_ROOT_TESTS = frozenset({"test_development_policy_guard.py"})
+
+RPA_001_FORBIDDEN_PATTERNS = (
+    re.compile(r"\brpa_helpers\b"),
+    re.compile(r"\binject_account_manager_scripts_path\b"),
+    re.compile(r"\bget_account_credential\b"),
+)
+
+RPA_002_FFMPEG_PATTERNS = (
+    re.compile(
+        r"\bsubprocess\.(run|call|Popen|check_output|check_call)\s*\([^)]*ffmpeg",
+        re.IGNORECASE | re.DOTALL,
+    ),
+    re.compile(
+        r"\bsubprocess\.(run|call|Popen|check_output|check_call)\(\s*['\"][^'\"]*ffmpeg",
+        re.IGNORECASE,
+    ),
+    re.compile(r"\bos\.system\s*\([^)]*ffmpeg", re.IGNORECASE | re.DOTALL),
+    re.compile(r"\bos\.popen\s*\([^)]*ffmpeg", re.IGNORECASE | re.DOTALL),
+    re.compile(
+        r"\bsubprocess\.(run|call|Popen|check_output|check_call)\([^)]*shell\s*=\s*True[^)]*ffmpeg",
+        re.IGNORECASE | re.DOTALL,
+    ),
+)
+
+RPA_002_EXEMPT_REL = "scripts/service/task_run_support.py"
+
+
+def _policy_msg(policy_id: str, source: str, detail: str) -> str:
+    return f"{policy_id} [{source}]: {detail}"
+
+
+def _rel(skill_root: str, path: str) -> str:
+    return os.path.relpath(path, skill_root).replace("\\", "/")
+
+
+def _should_skip_dirname(dirname: str) -> bool:
+    return dirname in SKIP_DIR_NAMES
+
+
+def _should_skip_rel_path(skill_root: str, abs_path: str) -> bool:
+    rel = _rel(skill_root, abs_path).replace("\\", "/")
+    for prefix in SKIP_REL_PREFIXES:
+        normalized = prefix.rstrip("/")
+        if rel == normalized or rel.startswith(normalized + "/"):
+            return True
+    return False
+
+
+def _walk_files(skill_root: str, rel_root: str, *, suffix: str | None = None) -> list[str]:
+    base = os.path.join(skill_root, rel_root)
+    if not os.path.exists(base):
+        return []
+    found: list[str] = []
+    for root, dirnames, filenames in os.walk(base):
+        kept: list[str] = []
+        for d in dirnames:
+            if _should_skip_dirname(d):
+                continue
+            sub = os.path.join(root, d)
+            if _should_skip_rel_path(skill_root, sub):
+                continue
+            kept.append(d)
+        dirnames[:] = kept
+        for name in filenames:
+            full = os.path.join(root, name)
+            if _should_skip_rel_path(skill_root, full):
+                continue
+            if suffix and not name.endswith(suffix):
+                continue
+            found.append(full)
+    return sorted(found)
+
+
+def _read_text(path: str) -> str:
+    with open(path, encoding="utf-8") as f:
+        return f.read()
+
+
+def _requirement_dependency_lines(req_text: str) -> list[str]:
+    return [
+        line.strip()
+        for line in req_text.splitlines()
+        if line.strip() and not line.strip().startswith("#")
+    ]
+
+
+def _scan_install_delivery_files(skill_root: str) -> list[str]:
+    offenders: list[str] = []
+    scan_specs: list[tuple[str, str | None]] = [
+        ("scripts", ".py"),
+        (".", ".ps1"),
+        (".", ".sh"),
+        (".", None),
+        (".github/workflows", ".yml"),
+        (".github/workflows", ".yaml"),
+    ]
+    seen: set[str] = set()
+    for rel_root, suffix in scan_specs:
+        if rel_root == "." and suffix is None:
+            req = os.path.join(skill_root, "requirements.txt")
+            if os.path.isfile(req):
+                seen.add(req)
+                text = _read_text(req)
+                for pattern in INSTALL_FORBIDDEN_PATTERNS:
+                    if pattern.search(text):
+                        offenders.append(
+                            _policy_msg(
+                                "POLICY-INSTALL-001",
+                                "development/RUNTIME.md; development/RPA.md §1.4; development/DEVELOPMENT.md §3.1",
+                                f"{_rel(skill_root, req)} matches {pattern.pattern}",
+                            )
+                        )
+            continue
+        for path in _walk_files(skill_root, rel_root, suffix=suffix):
+            if path in seen:
+                continue
+            seen.add(path)
+            if suffix == ".ps1" and not path.endswith(".ps1"):
+                continue
+            if suffix == ".sh" and not path.endswith(".sh"):
+                continue
+            text = _read_text(path)
+            for pattern in INSTALL_FORBIDDEN_PATTERNS:
+                if pattern.search(text):
+                    offenders.append(
+                        _policy_msg(
+                            "POLICY-INSTALL-001",
+                            "development/RUNTIME.md; development/RPA.md §1.4; development/DEVELOPMENT.md §3.1",
+                            f"{_rel(skill_root, path)} matches {pattern.pattern}",
+                        )
+                    )
+    return offenders
+
+
+def _gitignore_has_pattern(lines: list[str], pattern: str) -> bool:
+    normalized = [ln.strip() for ln in lines if ln.strip() and not ln.strip().startswith("#")]
+    return pattern in normalized
+
+
+class TestPolicyStructure001(unittest.TestCase):
+    def test_standard_scripts_layout_exists(self) -> None:
+        skill_root = get_skill_root()
+        missing: list[str] = []
+        for rel in STRUCTURE_PATHS:
+            path = os.path.join(skill_root, *rel.split("/"))
+            if not os.path.exists(path):
+                missing.append(rel)
+        self.assertEqual(
+            missing,
+            [],
+            msg=_policy_msg(
+                "POLICY-STRUCTURE-001",
+                "development/DEVELOPMENT.md §2; development/RUNTIME.md §目录结构",
+                "missing: " + ", ".join(missing),
+            ),
+        )
+
+
+class TestPolicyRuntime001(unittest.TestCase):
+    def test_no_vendored_jiangchang_skill_core(self) -> None:
+        skill_root = get_skill_root()
+        vendored = os.path.join(skill_root, "scripts", "jiangchang_skill_core")
+        self.assertFalse(
+            os.path.isdir(vendored),
+            msg=_policy_msg(
+                "POLICY-RUNTIME-001",
+                "development/RUNTIME.md §共享 Python Runtime",
+                _rel(skill_root, vendored),
+            ),
+        )
+
+
+class TestPolicyRuntime002(unittest.TestCase):
+    def test_requirements_txt_excludes_shared_runtime_deps(self) -> None:
+        skill_root = get_skill_root()
+        req_path = os.path.join(skill_root, "requirements.txt")
+        self.assertTrue(os.path.isfile(req_path), msg="requirements.txt missing")
+        text = _read_text(req_path)
+        dep_lines = _requirement_dependency_lines(text)
+        offenders: list[str] = []
+        for line in dep_lines:
+            lower = line.lower()
+            if "jiangchang-platform-kit" in lower:
+                offenders.append(line)
+            if re.search(r"(?<![\w.-])playwright(?![\w.-])", lower):
+                offenders.append(line)
+        self.assertEqual(
+            offenders,
+            [],
+            msg=_policy_msg(
+                "POLICY-RUNTIME-002",
+                "development/RUNTIME.md; development/DEVELOPMENT.md §3.1",
+                "forbidden dependency lines: " + "; ".join(offenders),
+            ),
+        )
+
+
+class TestPolicyInstall001(unittest.TestCase):
+    def test_delivery_files_have_no_auto_install_commands(self) -> None:
+        offenders = _scan_install_delivery_files(get_skill_root())
+        self.assertEqual(offenders, [], msg="\n".join(offenders))
+
+
+class TestPolicyConfig001(unittest.TestCase):
+    def test_env_example_exists(self) -> None:
+        skill_root = get_skill_root()
+        path = os.path.join(skill_root, ".env.example")
+        self.assertTrue(
+            os.path.isfile(path),
+            msg=_policy_msg(
+                "POLICY-CONFIG-001",
+                "development/CONFIG.md",
+                ".env.example missing at repo root",
+            ),
+        )
+
+
+class TestPolicyConfig002(unittest.TestCase):
+    def test_gitignore_ignores_env_files(self) -> None:
+        skill_root = get_skill_root()
+        path = os.path.join(skill_root, ".gitignore")
+        self.assertTrue(os.path.isfile(path), msg=".gitignore missing")
+        lines = _read_text(path).splitlines()
+        missing: list[str] = []
+        if not _gitignore_has_pattern(lines, ".env"):
+            missing.append(".env")
+        if not _gitignore_has_pattern(lines, "*.env.local"):
+            missing.append("*.env.local")
+        self.assertEqual(
+            missing,
+            [],
+            msg=_policy_msg(
+                "POLICY-CONFIG-002",
+                "development/CONFIG.md §红线：敏感信息不进 .env",
+                "missing gitignore entries: " + ", ".join(missing),
+            ),
+        )
+
+
+class TestPolicyConfig003(unittest.TestCase):
+    def test_business_scripts_do_not_read_os_environ_directly(self) -> None:
+        skill_root = get_skill_root()
+        offenders: list[str] = []
+        for path in _walk_files(skill_root, "scripts", suffix=".py"):
+            rel = _rel(skill_root, path)
+            if rel.replace("\\", "/") in CONFIG_ENV_BOUNDARY_FILES:
+                continue
+            text = _read_text(path)
+            for lineno, line in enumerate(text.splitlines(), 1):
+                if line.strip().startswith("#"):
+                    continue
+                for pattern in CONFIG_ENV_FORBIDDEN_PATTERNS:
+                    if pattern.search(line):
+                        offenders.append(
+                            f"{rel}:{lineno}: {line.strip()} — use jiangchang_skill_core.config.get*"
+                        )
+        self.assertEqual(
+            offenders,
+            [],
+            msg=_policy_msg(
+                "POLICY-CONFIG-003",
+                "development/CONFIG.md; development/ADAPTER.md",
+                "\n".join(offenders),
+            ),
+        )
+
+
+class TestPolicyTesting001(unittest.TestCase):
+    def test_root_tests_do_not_hardcode_real_targets(self) -> None:
+        skill_root = get_skill_root()
+        tests_dir = os.path.join(skill_root, "tests")
+        offenders: list[str] = []
+        for name in sorted(os.listdir(tests_dir)):
+            if not (name.startswith("test_") and name.endswith(".py")):
+                continue
+            if name in TESTING_001_SKIP_ROOT_TESTS:
+                continue
+            if name in TESTING_REAL_TARGET_ALLOWED_ROOT_TESTS:
+                continue
+            path = os.path.join(tests_dir, name)
+            text = _read_text(path)
+            for lineno, line in enumerate(text.splitlines(), 1):
+                stripped = line.strip()
+                if stripped.startswith("#"):
+                    continue
+                for token in TESTING_001_FORBIDDEN_SUBSTRINGS:
+                    if token in line:
+                        offenders.append(f"{name}:{lineno}: {stripped}")
+                        break
+        self.assertEqual(
+            offenders,
+            [],
+            msg=_policy_msg(
+                "POLICY-TESTING-001",
+                "development/TESTING.md §默认必跑测试要做什么",
+                "\n".join(offenders),
+            ),
+        )
+
+
+class TestPolicyTesting002(unittest.TestCase):
+    def test_integration_python_tests_are_sample_only(self) -> None:
+        skill_root = get_skill_root()
+        integration_dir = os.path.join(skill_root, "tests", "integration")
+        if not os.path.isdir(integration_dir):
+            return
+        offenders: list[str] = []
+        for name in sorted(os.listdir(integration_dir)):
+            if not name.endswith(".py"):
+                continue
+            if name.endswith(".sample"):
+                continue
+            offenders.append(f"tests/integration/{name}")
+        self.assertEqual(
+            offenders,
+            [],
+            msg=_policy_msg(
+                "POLICY-TESTING-002",
+                "development/TESTING.md §7 真实联调测试的安全约束",
+                "non-sample integration tests: " + ", ".join(offenders),
+            ),
+        )
+
+
+class TestPolicyRpa001(unittest.TestCase):
+    def test_scripts_have_no_account_manager_internal_imports(self) -> None:
+        skill_root = get_skill_root()
+        offenders: list[str] = []
+        for path in _walk_files(skill_root, "scripts", suffix=".py"):
+            rel = _rel(skill_root, path)
+            text = _read_text(path)
+            for pattern in RPA_001_FORBIDDEN_PATTERNS:
+                if pattern.search(text):
+                    offenders.append(f"{rel}: {pattern.pattern}")
+        self.assertEqual(
+            offenders,
+            [],
+            msg=_policy_msg(
+                "POLICY-RPA-001",
+                "development/ADAPTER.md §兄弟依赖; development/RPA.md §1.7",
+                "\n".join(offenders),
+            ),
+        )
+
+
+class TestPolicyRpa002(unittest.TestCase):
+    def test_scripts_do_not_invoke_ffmpeg_directly(self) -> None:
+        skill_root = get_skill_root()
+        offenders: list[str] = []
+        for path in _walk_files(skill_root, "scripts", suffix=".py"):
+            rel = _rel(skill_root, path)
+            if rel.replace("\\", "/") == RPA_002_EXEMPT_REL:
+                continue
+            text = _read_text(path)
+            for pattern in RPA_002_FFMPEG_PATTERNS:
+                if pattern.search(text):
+                    offenders.append(f"{rel}: {pattern.pattern}")
+        self.assertEqual(
+            offenders,
+            [],
+            msg=_policy_msg(
+                "POLICY-RPA-002",
+                "development/RPA.md §5.3 录屏成片标准",
+                "\n".join(offenders),
+            ),
+        )
+
+
+class TestPolicyDocs001(unittest.TestCase):
+    def test_policy_matrix_exists_and_lists_policy_ids(self) -> None:
+        skill_root = get_skill_root()
+        matrix_path = os.path.join(skill_root, "development", "POLICY_MATRIX.md")
+        self.assertTrue(
+            os.path.isfile(matrix_path),
+            msg=_policy_msg(
+                "POLICY-DOCS-001",
+                "development/POLICY_MATRIX.md",
+                "file missing",
+            ),
+        )
+        text = _read_text(matrix_path)
+        missing = [pid for pid in POLICY_IDS if pid not in text]
+        self.assertEqual(
+            missing,
+            [],
+            msg=_policy_msg(
+                "POLICY-DOCS-001",
+                "development/POLICY_MATRIX.md",
+                "missing policy_id entries: " + ", ".join(missing),
+            ),
+        )
+
+
+if __name__ == "__main__":
+    unittest.main()

tests/test_docs_standards.py

@@ -293,13 +293,13 @@ class TestDocsStandards(unittest.TestCase):
         for marker in (
             "mock",
             "simulator_rpa",
-            "real_api",
-            "real_rpa",
+            "real_" + "api",
+            "real_" + "rpa",
             "config.get",
             "sibling_bridge",
         ):
             self.assertIn(marker, text, msg=f"ADAPTER.md missing {marker!r}")
-        self.assertNotIn("ALLOW_REAL_API", text)
+        self.assertNotIn("ALLOW_" + "REAL_" + "API", text)
         self.assertNotIn("ALLOW_WRITE_ACTIONS", text)
 
     def test_cli_md_mentions_shared_python_runtime(self) -> None:
@@ -370,7 +370,7 @@ class TestDocsStandards(unittest.TestCase):
 
     def test_examples_top_level_modes_present(self) -> None:
         examples_dir = os.path.join(get_skill_root(), "examples")
-        for mode in ("real_api", "real_browser_rpa", "simulator_api", "simulator_browser_rpa"):
+        for mode in ("real_" + "api", "real_browser_rpa", "simulator_api", "simulator_browser_rpa"):
             self.assertTrue(
                 os.path.isdir(os.path.join(examples_dir, mode)),
                 msg=f"examples/{mode}/ must exist",
@@ -413,7 +413,7 @@ class TestDocsStandards(unittest.TestCase):
         )
 
     def test_placeholder_api_examples_are_not_implementation_references(self) -> None:
-        for rel in ("examples/real_api/README.md", "examples/simulator_api/README.md"):
+        for rel in (f"examples/{'real_' + 'api'}/README.md", "examples/simulator_api/README.md"):
             text = self._read(rel)
             self.assertTrue(
                 "占位" in text or "规划" in text,

tests/test_video_service.py

@@ -21,7 +21,7 @@ class TestEnvExampleVideoDefaults(unittest.TestCase):
         with open(path, encoding="utf-8") as f:
             text = f.read()
         for key in (
-            "OPENCLAW_TEST_TARGET=mock",
+            "OPENCLAW_" + "TEST_TARGET=mock",
             "OPENCLAW_RECORD_VIDEO=1",
             "OPENCLAW_ARTIFACTS_ON_FAILURE=1",
             "OPENCLAW_BROWSER_HEADLESS=0",