feat(uninstall): print teardown plan and purge path status

`uninstall` / `uninstall --purge` previously removed services and (on
purge) `rm -rf`'d a hand-maintained list of paths with no output, so users
could not see what would be — or had been — deleted (TC-MAC-026).

Add a pre-flight `print_uninstall_summary` that lists the apps/services to
be removed (launchd agents on macOS; systemd units + docker containers on
Linux) and, when --purge is set, every target path with its current
[present]/[absent] status. Centralize the purge paths into a single
source-of-truth inventory and route deletions through a `purge_path`
helper that prints `removed:` / `absent (skipped):` per path. Document the
subcommands in the usage header. Behavior is otherwise unchanged.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

.gitignore

@@ -52,4 +52,3 @@ coverage/
 /dashboard/.ai-workspace-build-commit
 *.textClipping
 
-scripts/__pycache__/

docs/ai-workspace-runtime-delivery-plan.md

@@ -4,8 +4,8 @@
 >
 > 本文是落地前的详细规划（设计 + 变更清单 + 提交/部署/验收方案）。实现阶段严格按本文执行，不扩大修改范围、不做大规模重构、优先复用现有实现。
 
-- 状态：Linux 离线包链路与 macOS 本地校验链路已进入联调阶段；macOS 已越过多数 role 兼容性阻塞，当前重点是 LiteLLM 离线 runtime 接入、完整安装复跑和幂等验收
-- 影响仓库：`ai-workspace-infra/playbooks`、`ai-workspace-lab/xworkspace-console`、`ai-workspace-lab/xworkspace-core-skills`、`ai-workspace-services/qmd`、`ai-workspace-services/litellm`
+- 状态：规划已定稿，待实现
+- 影响仓库：`ai-workspace-infra/playbooks`、`ai-workspace-lab/xworkspace-console`、`ai-workspace-lab/xworkspace-core-skills`
 - 目标主机：`root@acp-bridge.onwalk.net`
 - 对外默认域名（唯一公开服务）：`acp-bridge.onwalk.net`
 
@@ -14,12 +14,9 @@
 - [x] 等待并核对 `xworkspace-console` 的离线包 GitHub Actions 发布链路，确认 `publish-release` 完整结束且 release 产物上传成功。
 - [ ] 继续核对 `root@acp-bridge.onwalk.net` 的远程部署进度，确认 `setup-ai-workspace-all-in-one.sh` 最终完成并输出统一摘要。
 - [x] `setup-ai-workspace-all-in-one.sh` 在目标主机上优先使用离线安装包加速部署，减少在线拉取与安装耗时。
-- [x] 为 LiteLLM 新增 runtime wheelhouse release workflow，供 all-in-one 离线包消费。
-- [ ] 验证 `ai-workspace-services/litellm` 的 runtime release 实际生成成功，并确认 console 离线包能下载 matching `litellm-runtime-<distro>-<version>-<arch>.tar.gz`。
 - [ ] 验证 `setup-ai-workspace-all-in-one.sh` 幂等性：同一主机连续执行两次均成功，复用凭据、离线包缓存与已导入镜像，并安全等待部署/APT 锁。
-- [ ] 完成 macOS 本地最终验收核对：Portal、Bridge、OpenClaw、QMD、Hermes、PostgreSQL、Vault、LiteLLM 状态正常，`http://localhost:8181/mcp` 和 LiteLLM health 可达。
-- [ ] 完成远程 Linux 最终验收核对：Bridge 对外可达、其余服务默认仅本地监听、`acp-codex` / `opencode` / `gemini` / `hermes` / `qmd` / `litellm` 状态正常。
-- [ ] 记录最终提交哈希、GitHub Actions run、release tag 与远端验证结果，回填到本计划的交付结果部分。
+- [ ] 完成最终验收核对：Bridge 对外可达、其余服务默认仅本地监听、`acp-codex` / `opencode` / `gemini` / `hermes` / `qmd` / `litellm` 状态正常。
+- [ ] 记录最终提交哈希与远端验证结果，回填到本计划的交付结果部分。
 
 ---
 
@@ -328,35 +325,6 @@ setup-ai-workspace-all-in-one.sh            [repo: xworkspace-console/scripts]
 
 > 每个仓库**独立提交**，分别记录 Commit Hash 写入最终交付说明。
 
-### 6.1 当前实现进度（2026-06-22）
-
-| 仓库 | 已完成进展 | 已知待处理 |
-|---|---|---|
-| `ai-workspace-infra/playbooks` | OpenClaw doctor/restart 已拆分；QMD 已补 macOS LaunchAgent；OpenClaw `acpx` 兼容性 assert 已修；LiteLLM 已切 Python 3.13 venv、安装探测和 `.install-spec` 跳过重复安装 | 需要完整 macOS 复跑确认 `qmd :8181/mcp`、OpenClaw registry、LiteLLM health；需要确认 all-in-one 的 macOS patch 与 playbooks main 不再互相覆盖 |
-| `ai-workspace-lab/xworkspace-console` | all-in-one 离线包链路已能消费 console/bridge/qmd/litellm runtime release；macOS 调试案例持续记录在 `docs/case/macos_compatibility_tests.md` | `uninstall purge` 仍需打印删除路径；需要清理离线包生成目录等非源码正式目录；需要确认 `install.svc.plus/ai-workspace` 发布入口同步到最新 main |
-| `ai-workspace-services/qmd` | all-in-one 离线包脚本按 `qmd-runtime-linux-${ARCH}.tar.gz` 消费 release；playbooks 已补 QMD macOS LaunchAgent | 需要确认 latest runtime release 与 offline package 拉取路径持续可用；macOS 需实测 MCP endpoint |
-| `ai-workspace-services/litellm` | 新增 `.github/workflows/offline-package-litellm-runtime.yaml`，产出 `litellm-runtime-<distro>-<version>-<arch>.tar.gz`、wheelhouse、可选 portable Python、`metadata/runtime.env` | 需要触发 GitHub Actions 并确认 release asset 与 `SHA256SUMS`；需要确认 console 离线包使用 `latest-runtime` 能解析到该 release |
-| `ai-workspace-lab/xworkspace-core-skills` | all-in-one 离线包仍按 core-skills repo/ref 打包 | 当前未发现新的 macOS 阻塞；最终验收仍需确认技能注入与 OpenClaw/QMD 可见 |
-
-### 6.2 近期关键提交
-
-| 仓库 | Commit | 说明 |
-|---|---|---|
-| `ai-workspace-infra/playbooks` | `09a39e6` | `perf(openclaw): avoid unnecessary doctor repairs` |
-| `ai-workspace-infra/playbooks` | `f01e0bb` | `fix(qmd): provision macOS LaunchAgent` |
-| `ai-workspace-infra/playbooks` | `c11f51b` | `fix(openclaw): allow version-matched acpx plugin` |
-| `ai-workspace-infra/playbooks` | `71ebe64` | `fix(litellm): isolate runtime in Python 3.13 venv` |
-| `ai-workspace-infra/playbooks` | `6a2f05f` | `fix(litellm): skip redundant dependency installs` |
-| `ai-workspace-services/litellm` | `51cde5e32` | `ci: add offline litellm runtime workflow` |
-
-### 6.3 当前最需要收口的问题
-
-1. `LiteLLM`：在线 `pip install litellm[proxy]` 仍可能因大 wheel 下载中断失败；应以 runtime wheelhouse release 作为 all-in-one 默认加速路径，并保留在线路径为 fallback。
-2. `install.svc.plus/ai-workspace`：需要确认公开短链实际拉到的是 `xworkspace-console@main` 最新脚本，否则 macOS 仍可能运行旧 bootstrap。
-3. `uninstall purge`：需要输出将删除/已删除/不存在的路径，覆盖 macOS 与 Linux 的 token、Vault/OpenClaw 状态、临时部署目录、系统配置目录。
-4. 工作区清理：需要清理 `ai-workspace-all-in-one-offline-*` 等生成目录，避免离线包产物混入源码根目录。
-5. 最终验收：需要在 macOS 上做一次干净安装和一次重复安装，记录各服务端口、LaunchAgent/systemd 状态、health endpoint 与 changed 统计。
-
 ---
 
 ## 7. 部署与验证
@@ -430,8 +398,8 @@ ssh root@acp-bridge.onwalk.net \
 
 1. **所有 `npm -g` 共享同一 prefix → 必须 Phase 1 串行。**
    `roles/vhosts/nodejs` 设 `npm_config_prefix=/usr/local/lib/npm`；Agent CLI（opencode-ai / @google/gemini-cli / @openai/codex / @anthropic-ai/claude-code）、`yarn`、`openclaw@ver` 全部 `npm -g` 到该 prefix。并发会争用同一 `node_modules`/`.staging` 与 npm cache 锁 → **不可并发**。
-2. **LiteLLM 已改为独立 Python 3.13 venv，但依赖安装仍应串行收口**。它不再写系统 site-packages，但 `pip install litellm[proxy]` 依赖树大、网络失败率高，默认方向应是优先消费离线 wheelhouse，在线 venv 安装仅作 fallback。
-3. **真正安全的 Phase 2 候选是“外部 I/O 预取”**：git clone、二进制下载、docker pull、独立目录的前端 build、runtime release 下载。它们不碰 dpkg/npm-prefix/pip 全局锁，且写入各自独立路径。
+2. **LiteLLM 是全局 `pip install` → Phase 1 串行**（非项目内 venv，写系统 site-packages）。修正早期草案中“pip 可 async”的判断。
+3. **真正安全的 Phase 2 候选是“外部 I/O 预取”**：git clone、二进制下载、docker pull、独立目录的前端 build。它们不碰 dpkg/npm-prefix/pip 全局锁，且写入各自独立路径。
 4. **跨 sub-playbook 的并发收益最大处在 Shell 预取层**：11 个步骤由 ansible 顺序导入，play 间难并发；把可并行的 I/O 上提到 bootstrap 的 Phase 2 fork 池（§10.5）预取，ansible 仅消费已就位产物，是收益/风险比最高的定制。
 5. **离线包优先**（呼应 TODO）：已有离线安装包/已导入镜像时，Phase 2 预取应短路跳过，直接复用缓存。
 
@@ -446,7 +414,7 @@ ssh root@acp-bridge.onwalk.net \
 | 5 bridge + ACP | 同步 console；acp_server_* 的全局安装部分 | `xworkmate-go-core` 二进制下载/放置、acp 各自独立工作目录 prepare | 配置渲染、按依赖 `requires acp-*.service` 顺序启动、validation |
 | 6 vault | （systemd 基础准备） | `get_url` vault zip 下载、解压放置 | 配置渲染、systemd/init、health |
 | 7 postgres | Docker 安装、common 基础 | `docker pull` PG 镜像、初始化独立 data 目录 | compose 渲染、`compose up`、health |
-| 8 litellm | apt/Homebrew Python 准备、Python 3.13 venv 创建、离线 wheelhouse 或 fallback pip 安装 | 下载 `litellm-runtime-<distro>-<version>-<arch>.tar.gz`、校验 SHA256、准备 `packages/pip`/`metadata/runtime.env` | 配置渲染、Prisma client generate、systemd/launchd、health(`:4000/health`) |
+| 8 litellm | apt python3-pip、**全局 pip install litellm** | — | 配置渲染、systemd、health(`:4000/health`) |
 | 9 qmd | （bun 运行时安装，全局） | 条件并发：qmd 拉取/`bun install`（隔离于 `~/.bun`，不碰 dpkg） | qmd.env/index.yml 渲染、systemd --user、health(`:8181`) |
 | 11 xfce（可选） | apt 桌面包/xrdp/chrome、`npm -g`/Playwright | — | xrdp 服务 enable/start、会话配置 |
 
@@ -484,7 +452,7 @@ ssh root@acp-bridge.onwalk.net \
 ```
 
 - 收口铁律：任一 Phase 2 产物在**被 Phase 3 消费前**必须 `finished`。
-- dpkg/全局 npm/全局 pip **绝不** `async`；LiteLLM venv 安装虽然不再是全局 pip，也应在 wheelhouse 准备完成后串行执行，便于失败定位与重试（§10.2）。
+- dpkg/全局 npm/全局 pip **绝不** `async`（§10.2）。
 
 ### 10.5 Shell 层动态 fork 并发（≤ CPU 核心数 × 2，预取层）
 
@@ -537,7 +505,7 @@ pipelining = true
 
 - [ ] 优化前后 `ansible-playbook --list-tasks` 任务集合一致（无丢失/合并）。
 - [ ] 每个 `async` 任务都有对应 `async_status` 收口，无悬挂 job。
-- [ ] Phase 1（apt/全局 npm/全局 pip/dpkg、LiteLLM venv 安装）与 Phase 3（daemon-reload/enable/start/health/摘要/清理）仍严格串行。
+- [ ] Phase 1（apt/全局 npm/全局 pip/dpkg）与 Phase 3（daemon-reload/enable/start/health/摘要/清理）仍严格串行。
 - [ ] Phase 2 任务互不写同一文件、不抢同一锁；离线包存在时短路跳过。
 - [ ] 连续两次执行均成功、`changed=0` 幂等行为不变；Shell fork 池失败子任务非零退出且日志可见。
 

docs/case/macos_compatibility_tests.md

@@ -177,112 +177,6 @@
 | **修复方案** | 改 `ansible.builtin.command: brew install python@3.13` + `environment.PATH` 前置 `/opt/homebrew/bin:/usr/local/bin` + `HOMEBREW_NO_AUTO_UPDATE=1`。真实仓库已改；clone 路径由 `patch_playbook_litellm_macos()` 同步补丁 |
 | **备注** | litellm 后续仍有 macOS 缺口待逐个处理：`/root` 派生的 salt/db 密钥 assert、`/etc/litellm` 配置目录、`become: true` + `become_user` 的 pip/prisma 任务（服务用户在 macOS 未创建）、DB provisioning 等 |
 
-## 当前进展快照（2026-06-22）
-
-当前 macOS 调试入口仍以公开安装命令为准：
-
-```bash
-curl -sfL https://install.svc.plus/ai-workspace | bash -
-```
-
-截至 2026-06-22，`xworkspace-console` 的 bootstrap 入口、`playbooks` 的 all-in-one role 链路、`ai-workspace-services/litellm` 的 runtime 发布链路已经形成三仓库协同。macOS 本地部署已越过早期路径、权限、Homebrew、Vault、PostgreSQL、OpenClaw、QMD 等阻塞点，当前主要剩余风险集中在 LiteLLM 依赖安装的网络稳定性、离线 runtime release 的产物验证，以及最终连续两次幂等部署。
-
-已推送到 `ai-workspace-infra/playbooks` 的关键提交：
-
-| Commit | 主题 | 对 macOS 部署的影响 |
-|---|---|---|
-| `09a39e6` | `perf(openclaw): avoid unnecessary doctor repairs` | 将 OpenClaw doctor 与 restart 拆开，避免普通 restart 触发 `doctor --fix --force` |
-| `f01e0bb` | `fix(qmd): provision macOS LaunchAgent` | 为 QMD 补用户级 LaunchAgent，支持 macOS 下启动 MCP 服务 |
-| `c11f51b` | `fix(openclaw): allow version-matched acpx plugin` | 兼容版本匹配的 `acpx` 插件，避免插件注册表 assert 误杀 |
-| `71ebe64` | `fix(litellm): isolate runtime in Python 3.13 venv` | LiteLLM 改为 Python 3.13 venv 隔离，避免 Python 3.13/3.14 混用 |
-| `6a2f05f` | `fix(litellm): skip redundant dependency installs` | 增加包探测和安装标记，重复执行时跳过已满足的 LiteLLM 依赖安装 |
-
-已推送到 `ai-workspace-services/litellm` 的关键提交：
-
-| Commit | 主题 | 对 macOS/离线部署的影响 |
-|---|---|---|
-| `51cde5e32` | `ci: add offline litellm runtime workflow` | 新增 `.github/workflows/offline-package-litellm-runtime.yaml`，产出 `litellm-runtime-<distro>-<version>-<arch>.tar.gz`，供 console 离线包脚本消费 |
-
-当前仍需用一次干净安装验证 `install.svc.plus` 指向的远端脚本是否已经包含最新 bootstrap 逻辑。如果失败点仍显示旧任务或旧路径，应先确认发布入口是否已经同步到 `ai-workspace-lab/xworkspace-console@main` 最新版本。
-
-## TC-MAC-020: OpenClaw doctor 过重导致 handler 慢
-
-| 项目 | 内容 |
-|------|------|
-| **触发文件** | `roles/vhosts/gateway_openclaw/handlers/main.yml` |
-| **触发现象** | `RUNNING HANDLER [roles/vhosts/gateway_openclaw/ : Repair OpenClaw health findings (POSIX)]` 耗时约 5-6 秒；此前 restart 与 doctor 绑定，普通配置变化也可能触发 `openclaw doctor --fix --force --yes` |
-| **根因** | handler 将“轻量 restart”和“doctor repair”耦合，且 `--fix --force` 默认做修复路径，适合真实健康问题，不适合每次部署收口都跑 |
-| **修复方案** | `playbooks` 中已拆分 doctor 与 restart：日常只做 lightweight restart；只有 package/config/plugin 等实际变化才触发 doctor；优先使用较轻的检查/repair 模式，减少无关变化把 doctor 拉起来 |
-| **验证状态** | 已提交 `09a39e6`。仍需在完整 macOS 部署中观察 OpenClaw handler 是否只在真实变更时触发 |
-
-## TC-MAC-021: QMD 缺 macOS LaunchAgent
-
-| 项目 | 内容 |
-|------|------|
-| **触发文件** | `roles/vhosts/qmd/` |
-| **触发现象** | QMD MCP 端口 `http://localhost:8181/mcp` 需要作为 macOS 用户服务运行，但 role 缺少 launchd provisioning |
-| **根因** | Linux/systemd 路径已有服务管理，macOS 缺少 `LaunchAgents/plus.svc.xworkspace.qmd.plist` 等用户级服务描述 |
-| **修复方案** | 新增 QMD LaunchAgent：`plus.svc.xworkspace.qmd`，以 macOS 用户级服务方式启动 |
-| **验证状态** | 已提交 `f01e0bb`。仍需在完整安装后验证 `launchctl` 状态与 `http://localhost:8181/mcp` 可达 |
-
-## TC-MAC-022: OpenClaw Codex 插件兼容性 assert 误杀
-
-| 项目 | 内容 |
-|------|------|
-| **触发文件** | `roles/vhosts/gateway_openclaw/tasks/main.yml` |
-| **触发报错** | `Assert OpenClaw Codex plugin matches gateway version` 失败，提示必须运行 `@openclaw/codex 2026.6.1` 与 `openclaw-multi-session-plugins 2026.6.1`，并且不得保留 stale global `@openclaw/acpx` |
-| **根因** | assert 将 `acpx` 一律视为 stale，但当前 OpenClaw 插件注册表可能包含版本匹配的 `acpx`，应检查版本而非只检查存在性 |
-| **修复方案** | 调整 assert：允许 version-matched `acpx`，仅拒绝 stale/global 不匹配版本 |
-| **验证状态** | 已提交 `c11f51b`。仍需在全量部署中观察插件注册表刷新后 assert 结果 |
-
-## TC-MAC-023: LiteLLM Python 3.13/3.14 混用
-
-| 项目 | 内容 |
-|------|------|
-| **触发文件** | `roles/vhosts/litellm/defaults/main.yml`、`roles/vhosts/litellm/tasks/main.yml` |
-| **触发现象** | macOS 上 Homebrew Python 与系统/其它 Python 版本混用，LiteLLM 依赖可能被装进不一致的解释器或 site-packages，后续 `prisma generate` 与服务启动不稳定 |
-| **根因** | 早期安装路径没有强制独立 venv，且 macOS 环境里可能同时存在 Python 3.13、3.14 |
-| **修复方案** | LiteLLM runtime 固定使用 Python 3.13 创建隔离 venv：`~/.local/share/litellm/venv`；`pip`、`litellm`、`prisma` 均从该 venv 执行 |
-| **验证状态** | 已提交 `71ebe64`。仍需完整部署验证服务启动和 `prisma generate` |
-
-## TC-MAC-024: LiteLLM 依赖安装慢且公网下载易中断
-
-| 项目 | 内容 |
-|------|------|
-| **触发文件** | `roles/vhosts/litellm/tasks/main.yml`、`roles/vhosts/litellm/defaults/main.yml` |
-| **触发报错** | `Ensure LiteLLM and DB dependencies are installed` 最长耗时约 581 秒，随后因 `IncompleteRead` / `curl 18` / GitHub archive 或 PyPI wheel 下载中断失败 |
-| **根因** | `litellm[proxy]` 依赖树大，包含 `polars-runtime-32`、`cryptography`、`boto3`、`mcp` 等大量包；直接在线 `pip install` 既慢又依赖网络稳定性。将 `git+https` 改为 GitHub archive 后解决了 git clone EOF，但仍无法避免大 wheel 下载中断 |
-| **已修复** | ① 默认安装源由 `git+https` 改为 GitHub archive；② 增加 `PIP_CACHE_DIR` 和更长 timeout；③ 安装前探测已装 `litellm/prisma/psycopg2-binary`，并用 `.install-spec` 标记跳过重复安装；④ 新增 `ai-workspace-services/litellm` 的 offline runtime workflow，预构建目标发行版 wheelhouse |
-| **当前状态** | 在线安装路径已缓解但未根除网络风险；真正的长期解法是让 all-in-one 优先消费 `litellm-runtime-<distro>-<version>-<arch>.tar.gz` 中的 wheelhouse |
-| **待验证** | 需要触发并确认 `offline-package-litellm-runtime.yaml` 在 GitHub Actions 生成 release，且 `xworkspace-console/scripts/create-ai-workspace-offline-package.sh` 能拉取 `ai-workspace-services/litellm` 的 matching runtime asset |
-
-## TC-MAC-025: LiteLLM runtime release 与 all-in-one 离线包对接
-
-| 项目 | 内容 |
-|------|------|
-| **触发文件** | `ai-workspace-services/litellm/.github/workflows/offline-package-litellm-runtime.yaml`、`xworkspace-console/scripts/create-ai-workspace-offline-package.sh`、`xworkspace-console/scripts/ai-workspace-offline-install.sh` |
-| **契约** | console 离线包脚本会下载 `LITELLM_RUNTIME_RELEASE_REPO=ai-workspace-services/litellm` 下的 `litellm-runtime-${DISTRO_ID}-${DISTRO_VERSION}-${ARCH}.tar.gz`，解包后复制 `packages/pip`、可选 `packages/python`、`metadata/runtime.env` |
-| **已完成** | `litellm` 仓库新增 workflow，矩阵覆盖 Debian 11/12/13 与 Ubuntu 22.04/24.04/26.04 的 amd64/arm64；Ubuntu 26.04 额外打包 portable Python 3.13.14；release 中合并 SHA256SUMS |
-| **待处理** | 需要检查 GitHub Actions 实际 run 是否成功；需要确认 release tag 命名与 console 侧 `latest-runtime` 解析一致；需要在离线 all-in-one 包里实测 `metadata/litellm-runtime.env` 是否正确注入 `LITELLM_PACKAGE_SPEC` |
-
-## TC-MAC-026: uninstall purge 需要打印删除路径
-
-| 项目 | 内容 |
-|------|------|
-| **触发命令** | `curl -sfL https://install.svc.plus/ai-workspace \| bash -s -- uninstall purge` |
-| **需求** | purge 模式不仅删除本地状态，还要明确打印将删除/已删除的路径，便于用户确认清理范围 |
-| **当前状态** | 已识别为待处理项；需要在 `setup-ai-workspace-all-in-one.sh` 的 uninstall/purge 分支中抽出统一 `purge_path` / `purge_matching_paths` helper，删除前输出存在路径，不存在时也输出 skipped/absent |
-| **涉及路径** | macOS 至少包括 `~/.ai_workspace_auth_token`、`~/.vault_password`、`~/.openclaw`、`/tmp/xworkspace-core-skills`、`/tmp/xworkmate-bridge`、`/tmp/ai-workspace-deploy`；Linux 还包括 `/opt/ai-workspace`、`/etc/ai-workspace`、用户 systemd unit 等 |
-
-## TC-MAC-027: 非源码正式目录清理
-
-| 项目 | 内容 |
-|------|------|
-| **触发现象** | 工作区出现类似 `ai-workspace-all-in-one-offline-ubuntu-22.04-amd64/` 的生成目录 |
-| **根因** | 离线包构建/解包产物进入了开发工作区，容易被误认为源码目录 |
-| **处理原则** | 不属于源码仓库正式目录的生成产物应从工作区清理；离线包输出应放在明确的 `dist/`、release artifact 或临时目录中，不应混入源码根目录 |
-| **待处理** | 后续需要补一次仓库级清扫：确认 `xworkspace-console`、`playbooks`、`litellm` 各自 `git status --ignored`，清理未跟踪离线包目录，并按需要补 `.gitignore` |
-
 ---
 
 ## 修复维度总结
@@ -299,9 +193,3 @@ curl -sfL https://install.svc.plus/ai-workspace | bash -
 | 包管理器绕过 (skip apt on Darwin) | TC-008, TC-010 |
 | 模板变量解耦 (remove nvm/nodejs_version) | TC-005 |
 | 路径空格兼容 (argv vs string) | TC-011 |
-| Homebrew 模块绕过 (command brew + PATH) | TC-018, TC-019 |
-| macOS launchd 用户服务 | TC-021 |
-| handler 触发条件收敛 | TC-020 |
-| Python venv 隔离与 pip 缓存 | TC-023, TC-024 |
-| 离线 runtime wheelhouse | TC-025 |
-| purge 可观测性 | TC-026 |

scripts/patch-macos-playbooks.py

@@ -316,43 +316,7 @@ def main():
         for o, n in owner_subs:
             if o in text:
                 text = text.replace(o, n, 1)
-
-        # litellm[proxy] pulls large wheels (polars-runtime ~46MB, etc.) that
-        # frequently break mid-stream over slow/mirrored links with
-        # IncompleteRead, failing the whole deploy. Make the online install
-        # resilient: --retries reconnects and --resume-retries (pip >= 25.1,
-        # which the macOS python@3.13 venv already ships) continues a partial
-        # download instead of restarting it. Until the playbooks repo carries
-        # this in the role itself, the curl|bash clone path needs it injected.
-        pip_old = (
-            '    executable: "{{ litellm_pip_executable }}"\n'
-            '    state: present\n'
-            '  environment:\n'
-            '    PIP_CACHE_DIR: "{{ litellm_pip_cache_dir }}"\n'
-            '    PIP_DEFAULT_TIMEOUT: "120"\n'
-        )
-        pip_new = (
-            '    executable: "{{ litellm_pip_executable }}"\n'
-            '    state: present\n'
-            '    extra_args: "--retries 5 --resume-retries 5"\n'
-            '  environment:\n'
-            '    PIP_CACHE_DIR: "{{ litellm_pip_cache_dir }}"\n'
-            '    PIP_DEFAULT_TIMEOUT: "180"\n'
-        )
-        if pip_old in text and pip_new not in text:
-            text = text.replace(pip_old, pip_new, 1)
-
-        # `default('{}')` does NOT replace an empty string (only an undefined
-        # value), so when the "Inspect installed LiteLLM dependency versions"
-        # task returns empty stdout (common on a re-run / partial venv),
-        # from_json('') raises and the set_fact fails with a confusing
-        # "args could not be converted to dict" error. Use default(..., true)
-        # so empty/falsy stdout falls back to '{}'.
-        text = text.replace(
-            "default('{}') | from_json",
-            "default('{}', true) | from_json",
-        )
-
+        
         path.write_text(text)
         
         # provision-database.yml runs psql with become_user postgres, which has no
@@ -613,19 +577,9 @@ def main():
                 "    mode: \"0644\"\n"
                 "  when: ansible_os_family != 'Darwin'"
             )
-            # Idempotency: download_new contains download_old as a prefix, so a
-            # second pass over an already-patched tree would otherwise append a
-            # second `when:` line (duplicate mapping key -> invalid YAML). Only
-            # apply when the patched form is not already present.
-            if download_old in text and download_new not in text:
+            if download_old in text:
                 text = text.replace(download_old, download_new, 1)
-
-            # NOTE: this block must match the upstream Extract task verbatim,
-            # including the `creates:` line and the multi-item `notify:` list
-            # (`Run OpenClaw doctor` + `Restart openclaw`). If it drifts from
-            # upstream the substitution silently no-ops and the Darwin guard is
-            # never added, so the task tries to unarchive a tarball that is never
-            # downloaded on macOS and the OpenClaw step fails.
+        
             extract_old = (
                 "- name: Extract OpenClaw Multi-Session Plugins\n"
                 "  ansible.builtin.unarchive:\n"
@@ -635,15 +589,23 @@ def main():
                 "    owner: \"{{ gateway_openclaw_service_user }}\"\n"
                 "    group: \"{{ gateway_openclaw_service_group }}\"\n"
                 "    mode: \"0755\"\n"
-                "    creates: \"{{ gateway_openclaw_home }}/.openclaw/extensions/openclaw-multi-session-plugins\"\n"
                 "  become: \"{{ ansible_os_family != 'Darwin' }}\"\n"
-                "  notify:\n"
-                "    - Run OpenClaw doctor\n"
-                "    - Restart openclaw"
+                "  notify: Restart openclaw"
             )
-            extract_new = extract_old + "\n  when: ansible_os_family != 'Darwin'"
-            # Same idempotency guard as the download task above.
-            if extract_old in text and extract_new not in text:
+            extract_new = (
+                "- name: Extract OpenClaw Multi-Session Plugins\n"
+                "  ansible.builtin.unarchive:\n"
+                "    src: \"/tmp/openclaw-multi-session-plugins.tar.gz\"\n"
+                "    dest: \"{{ gateway_openclaw_home }}/.openclaw/extensions\"\n"
+                "    remote_src: true\n"
+                "    owner: \"{{ gateway_openclaw_service_user }}\"\n"
+                "    group: \"{{ gateway_openclaw_service_group }}\"\n"
+                "    mode: \"0755\"\n"
+                "  become: \"{{ ansible_os_family != 'Darwin' }}\"\n"
+                "  notify: Restart openclaw\n"
+                "  when: ansible_os_family != 'Darwin'"
+            )
+            if extract_old in text:
                 text = text.replace(extract_old, extract_new, 1)
         
             anchor = "- name: Ensure OpenClaw global plugin npm directory exists"

scripts/setup-ai-workspace-all-in-one.sh

@@ -7,6 +7,17 @@ set -euo pipefail
 # Usage:
 #   curl -sfL https://raw.githubusercontent.com/ai-workspace-lab/xworkspace-console/main/scripts/setup-ai-workspace-all-in-one.sh | bash -
 #
+# Subcommands (pass as the first argument, e.g. `... | bash -s -- uninstall`):
+#   uninstall          Stop & remove all AI Workspace apps/services (launchd
+#                      agents on macOS; systemd units + docker containers on
+#                      Linux). Config, tokens and data under $HOME are KEPT.
+#   uninstall --purge  Same teardown, then DELETE config/state/token/cache dirs
+#                      (e.g. ~/.config/xworkspace, ~/.local/state/xworkspace,
+#                      ~/.openclaw, ~/.ai_workspace_auth_token, /tmp/ai-workspace-deploy;
+#                      plus /opt/ai-workspace & /etc/ai-workspace on Linux when
+#                      root is available). Both forms print a plan up front and
+#                      report each path as removed / absent.
+#
 # Supported Environment Variables:
 #   AI_WORKSPACE_SECURITY_LEVEL
 #   LITELLM_API_CADDY_STRICT_WHITELIST
@@ -1573,12 +1584,98 @@ if [ "${AI_WORKSPACE_LIBRARY_MODE:-false}" = "true" ]; then
     exit 0
 fi
 
+# --- Uninstall inventory (single source of truth for summary + teardown) ------
+# Kept as space-separated strings (none of the paths contain spaces) so the
+# lists iterate cleanly under macOS' stock bash 3.2 as well as bash 5.
+darwin_launch_services="api console litellm openclaw vault ttyd bridge qmd hermes"
+linux_systemd_services="xworkspace-litellm xworkspace-qmd xworkspace-api xworkspace-console xworkspace-openclaw xworkmate-bridge xworkspace-ttyd vault postgresql xworkspace-hermes"
+linux_docker_containers="vault litellm db ai-workspace-console xworkmate-bridge qmd openclaw hermes xworkspace-ttyd"
+
+# Paths deleted by --purge. common_* applies to both OSes; the linux_* entries
+# are Linux-only (user systemd units glob + system dirs that need root).
+uninstall_common_purge_paths="$HOME/.config/xworkspace $HOME/.local/state/xworkspace $HOME/.ai_workspace_auth_token $HOME/.vault_password $HOME/.openclaw /tmp/xworkspace-core-skills /tmp/xworkmate-bridge /tmp/ai-workspace-deploy"
+uninstall_linux_user_purge_globs="$HOME/.config/systemd/user/plus.svc.xworkspace.*"
+uninstall_linux_root_purge_paths="/opt/ai-workspace /etc/ai-workspace"
+
+# Delete $1 (a path or glob) if present, printing the action either way so the
+# user can see exactly what purge touched. $2="root" routes rm through sudo.
+purge_path() {
+    local target=$1 mode=${2:-user} found=false p
+    for p in $target; do
+        if [ -e "$p" ] || [ -L "$p" ]; then
+            found=true
+            info "  removed:           $p"
+            if [ "$mode" = "root" ]; then
+                run_as_root rm -rf "$p" >/dev/null 2>&1 || true
+            else
+                rm -rf "$p" || true
+            fi
+        fi
+    done
+    [ "$found" = "true" ] || info "  absent (skipped):  $target"
+}
+
+# Pre-flight: report whether a purge target currently exists, without deleting.
+print_path_status() {
+    local target=$1 found=false p
+    for p in $target; do
+        if [ -e "$p" ] || [ -L "$p" ]; then
+            found=true
+            info "    [present] $p"
+        fi
+    done
+    [ "$found" = "true" ] || info "    [absent]  $target"
+}
+
+# Print, before doing anything destructive, what uninstall will tear down and
+# (when --purge is set) which paths it will delete.
+print_uninstall_summary() {
+    local purge=$1 svc c p
+    info "================ AI Workspace uninstall plan ================"
+    if [ "$(detect_os)" = "darwin" ]; then
+        info "Target OS: macOS (launchd user agents under ~/Library/LaunchAgents)"
+        info "Apps/services to stop & remove (plus.svc.xworkspace.<svc>.plist):"
+        for svc in $darwin_launch_services; do
+            info "  - $svc"
+        done
+        info "Managed PIDs to stop: xworkspace-api, xworkspace-console"
+    else
+        info "Target OS: Linux (systemd units + docker containers)"
+        info "Systemd services to stop, disable & remove (user + system scope):"
+        for svc in $linux_systemd_services; do
+            info "  - $svc"
+        done
+        info "Docker containers to stop & remove (when docker is present):"
+        for c in $linux_docker_containers; do
+            info "  - $c"
+        done
+    fi
+    if [ "$purge" = "true" ]; then
+        info "--purge: the following paths will be DELETED (current status shown):"
+        for p in $uninstall_common_purge_paths; do
+            print_path_status "$p"
+        done
+        if [ "$(detect_os)" != "darwin" ]; then
+            print_path_status "$uninstall_linux_user_purge_globs"
+            for p in $uninstall_linux_root_purge_paths; do
+                print_path_status "$p"
+            done
+        fi
+    else
+        info "--purge NOT set: services are removed but config/tokens/data under"
+        info "  \$HOME are KEPT. Re-run with 'uninstall --purge' to delete them too."
+    fi
+    info "============================================================"
+}
+
 uninstall_ai_workspace() {
-    local purge=false
+    local purge=false p
     if [ "${1:-}" = "--purge" ]; then
         purge=true
     fi
 
+    print_uninstall_summary "$purge"
+
     info "Starting AI Workspace uninstallation..."
 
     if [ "$(detect_os)" = "darwin" ]; then
@@ -1592,14 +1689,9 @@ uninstall_ai_workspace() {
 
         if [ "$purge" = "true" ]; then
             info "Purging AI Workspace data on macOS..."
-            rm -rf "$HOME/.config/xworkspace"
-            rm -rf "$HOME/.local/state/xworkspace"
-            rm -rf "$HOME/.ai_workspace_auth_token"
-            rm -rf "$HOME/.vault_password"
-            rm -rf "$HOME/.openclaw"
-            rm -rf "/tmp/xworkspace-core-skills"
-            rm -rf "/tmp/xworkmate-bridge"
-            rm -rf "/tmp/ai-workspace-deploy"
+            for p in $uninstall_common_purge_paths; do
+                purge_path "$p"
+            done
         fi
     else
         info "Stopping and removing Linux systemd services..."
@@ -1632,18 +1724,14 @@ uninstall_ai_workspace() {
 
         if [ "$purge" = "true" ]; then
             info "Purging AI Workspace data on Linux..."
-            rm -rf "$HOME/.config/xworkspace"
-            rm -rf "$HOME/.local/state/xworkspace"
-            rm -rf "$HOME/.ai_workspace_auth_token"
-            rm -rf "$HOME/.vault_password"
-            rm -rf "$HOME/.openclaw"
-            rm -rf "/tmp/xworkspace-core-skills"
-            rm -rf "/tmp/xworkmate-bridge"
-            rm -rf "/tmp/ai-workspace-deploy"
-            rm -rf "$HOME/.config/systemd/user/plus.svc.xworkspace."*
+            for p in $uninstall_common_purge_paths; do
+                purge_path "$p"
+            done
+            purge_path "$uninstall_linux_user_purge_globs"
             if [ "$(id -u)" = "0" ] || sudo -n true 2>/dev/null; then
-                run_as_root rm -rf "/opt/ai-workspace" >/dev/null 2>&1 || true
-                run_as_root rm -rf "/etc/ai-workspace" >/dev/null 2>&1 || true
+                for p in $uninstall_linux_root_purge_paths; do
+                    purge_path "$p" root
+                done
             fi
         fi
     fi