Feature/wsl worker

.github/workflows/quality.yml

@@ -23,7 +23,7 @@ jobs:
           sudo apt-get update
           sudo apt-get install -y libwebkit2gtk-4.1-dev libappindicator3-dev librsvg2-dev patchelf
       - uses: dtolnay/rust-toolchain@1.88.0
-      - run: cargo check --locked --manifest-path src-tauri/Cargo.toml --all-targets
+      - run: cargo check --locked --workspace --manifest-path src-tauri/Cargo.toml --all-targets
 
   frontend:
     runs-on: ubuntu-22.04
@@ -63,13 +63,13 @@ jobs:
         uses: docker://rhysd/actionlint:1.7.12
 
   shellcheck:
-    name: ShellCheck bundled WSL assets
+    name: ShellCheck WSL bootstrap
     runs-on: ubuntu-22.04
     steps:
       - uses: actions/checkout@v7
         with:
           ref: ${{ inputs.target_sha }}
-      - name: Check bundled WSL operation scripts
+      - name: Check WSL bootstrap script
         run: >-
           docker run --rm
           --volume "$PWD:/work"
@@ -92,7 +92,7 @@ jobs:
         shell: bash
         run: |
           mkdir -p ci-diagnostics
-          cargo fmt --manifest-path src-tauri/Cargo.toml -- --check 2>&1 | tee ci-diagnostics/fmt.log
+          cargo fmt --all --manifest-path src-tauri/Cargo.toml -- --check 2>&1 | tee ci-diagnostics/fmt.log
       - name: Upload Rust formatting diagnostics
         if: ${{ always() }}
         uses: actions/upload-artifact@v4
@@ -127,6 +127,13 @@ jobs:
       - uses: swatinem/rust-cache@v2
         with:
           workspaces: './src-tauri -> target'
+      - name: Prepare Windows Cargo resource placeholders
+        if: runner.os == 'Windows'
+        shell: bash
+        run: |
+          mkdir -p src-tauri/target/wsl-worker/current
+          printf 'quality-check-placeholder\n' > src-tauri/target/wsl-worker/current/worker
+          printf '{}\n' > src-tauri/target/wsl-worker/current/manifest.json
       - name: Capture runner diagnostics
         shell: bash
         run: |
@@ -143,11 +150,11 @@ jobs:
           } | tee ci-diagnostics/toolchain.txt
       - name: Check all Rust targets
         shell: bash
-        run: cargo check --locked --manifest-path src-tauri/Cargo.toml --all-targets 2>&1 | tee ci-diagnostics/check.log
+        run: cargo check --locked --workspace --manifest-path src-tauri/Cargo.toml --all-targets 2>&1 | tee ci-diagnostics/check.log
       - name: Lint all Rust targets
         if: ${{ always() }}
         shell: bash
-        run: cargo clippy --locked --manifest-path src-tauri/Cargo.toml --all-targets -- -D warnings 2>&1 | tee ci-diagnostics/clippy.log
+        run: cargo clippy --locked --workspace --manifest-path src-tauri/Cargo.toml --all-targets -- -D warnings 2>&1 | tee ci-diagnostics/clippy.log
       - name: Upload Rust static diagnostics
         if: ${{ always() }}
         uses: actions/upload-artifact@v4
@@ -171,6 +178,12 @@ jobs:
       - uses: actions/checkout@v7
         with:
           ref: ${{ inputs.target_sha }}
+      - uses: pnpm/action-setup@v6
+      - uses: actions/setup-node@v7
+        with:
+          node-version: lts/*
+          cache: pnpm
+      - run: pnpm install --frozen-lockfile
       - name: Install Tauri system dependencies
         if: runner.os == 'Linux'
         run: |
@@ -180,6 +193,13 @@ jobs:
       - uses: swatinem/rust-cache@v2
         with:
           workspaces: './src-tauri -> target'
+      - name: Prepare Windows Cargo resource placeholders
+        if: runner.os == 'Windows'
+        shell: bash
+        run: |
+          mkdir -p src-tauri/target/wsl-worker/current
+          printf 'quality-check-placeholder\n' > src-tauri/target/wsl-worker/current/worker
+          printf '{}\n' > src-tauri/target/wsl-worker/current/manifest.json
       - name: Capture runner diagnostics
         shell: bash
         run: |
@@ -196,7 +216,17 @@ jobs:
           } | tee ci-diagnostics/toolchain.txt
       - name: Run Rust tests
         shell: bash
-        run: cargo test --locked --manifest-path src-tauri/Cargo.toml 2>&1 | tee ci-diagnostics/test.log
+        run: cargo test --locked --workspace --manifest-path src-tauri/Cargo.toml 2>&1 | tee ci-diagnostics/test.log
+      - name: Verify CLI private installation updates
+        shell: bash
+        run: |
+          cargo test --locked --workspace --manifest-path src-tauri/Cargo.toml native_update_private_only_cli_installation_in_place -- --ignored --nocapture 2>&1 | tee ci-diagnostics/cli-private-update.log
+          grep -Eq 'test native_workflow_integration_support::update_lifecycle::native_update_private_only_cli_installation_in_place \.\.\. ok' ci-diagnostics/cli-private-update.log
+      - name: Verify CLI Eve directory updates
+        shell: bash
+        run: |
+          cargo test --locked --workspace --manifest-path src-tauri/Cargo.toml native_eve_update_restores_recorded_missing_installation -- --ignored --nocapture 2>&1 | tee ci-diagnostics/cli-eve-update.log
+          grep -Eq 'test native_workflow_integration_support::update_lifecycle::native_eve_update_restores_recorded_missing_installation \.\.\. ok' ci-diagnostics/cli-eve-update.log
       - name: Upload Rust test diagnostics
         if: ${{ always() }}
         uses: actions/upload-artifact@v4

.github/workflows/release.yml

@@ -68,6 +68,34 @@ jobs:
     with:
       target_sha: ${{ needs.validate.outputs.commit_sha }}
 
+  build-wsl-worker:
+    needs: [validate, quality]
+    runs-on: ubuntu-22.04
+    steps:
+      - uses: actions/checkout@v7
+        with:
+          ref: ${{ needs.validate.outputs.commit_sha }}
+      - uses: actions/setup-node@v7
+        with:
+          node-version: 22
+      - uses: dtolnay/rust-toolchain@stable
+        with:
+          targets: x86_64-unknown-linux-musl
+      - uses: swatinem/rust-cache@v2
+        with:
+          workspaces: "./src-tauri -> target"
+      - name: Build WSL Worker artifact
+        run: node scripts/prepare-wsl-worker.mjs
+      - name: Verify WSL Worker artifact
+        run: node scripts/prepare-wsl-worker.mjs --verify src-tauri/target/wsl-worker/current
+      - name: Upload WSL Worker artifact
+        uses: actions/upload-artifact@v4
+        with:
+          name: wsl-worker-${{ needs.validate.outputs.commit_sha }}
+          path: src-tauri/target/wsl-worker/current
+          if-no-files-found: error
+          retention-days: 7
+
   prepare-release:
     needs: [validate, quality]
     runs-on: ubuntu-22.04
@@ -184,7 +212,7 @@ jobs:
           echo "release_id=$RELEASE_ID" >> "$GITHUB_OUTPUT"
 
   build-release:
-    needs: [validate, quality, prepare-release]
+    needs: [validate, quality, prepare-release, build-wsl-worker]
     permissions:
       contents: write
     strategy:
@@ -229,6 +257,15 @@ jobs:
         with:
           workspaces: "./src-tauri -> target"
       - run: pnpm install --frozen-lockfile
+      - name: Download WSL Worker artifact
+        if: ${{ runner.os == 'Windows' }}
+        uses: actions/download-artifact@v4
+        with:
+          name: wsl-worker-${{ needs.validate.outputs.commit_sha }}
+          path: src-tauri/target/wsl-worker/current
+      - name: Verify WSL Worker artifact
+        if: ${{ runner.os == 'Windows' }}
+        run: pnpm verify:wsl-worker
       - name: Build and upload release bundles
         uses: tauri-apps/tauri-action@1deb371b0cd8bd54025b384f1cd735e725c4060f # v1.0.0
         env:
@@ -244,6 +281,12 @@ jobs:
           retryAttempts: 3
           releaseAssetNamePattern: skill-deck_[version]_${{ matrix.asset_platform }}_[arch][setup][ext]
           args: ${{ needs.validate.outputs.prerelease == 'true' && matrix.prerelease_args || matrix.args }}
+      - name: Verify NSIS WSL Worker resource
+        if: ${{ runner.os == 'Windows' }}
+        run: pwsh -NoProfile -File scripts/verify-windows-worker-bundle.ps1 -BundleKind nsis
+      - name: Verify MSI WSL Worker resource
+        if: ${{ runner.os == 'Windows' && needs.validate.outputs.prerelease != 'true' }}
+        run: pwsh -NoProfile -File scripts/verify-windows-worker-bundle.ps1 -BundleKind msi
 
   verify-release:
     needs: [validate, prepare-release, build-release]

CHANGELOG.md

@@ -7,6 +7,10 @@
 
 ## [Unreleased]
 
+### Changed
+
+- **简化 Skill 维护操作** — 来源信息不完整或上游已删除时，用户可以保留或移除当前 Skill；需要重新使用时，通过普通安装流程重新安装，不再提供独立的来源修复和重新安装操作。
+
 ## [1.7.1] - 2026-08-20
 
 ### Added

README.md

@@ -38,7 +38,7 @@
 ## ✨ Core capabilities
 
 - **Discover and install** — Browse available Skills, review their sources, documentation, and security information, or install from GitHub, Git, local directories, Well-known URLs, raw `SKILL.md` files, and ZIP/tar archives
-- **Browse and maintain** — Read installed Skill content, check and apply updates, and select a new source when the saved source no longer works; private GitHub sources can reuse local Git credentials for update checks
+- **Browse and maintain** — Read installed Skill content and check or apply updates; keep Skills whose sources are unavailable, or remove them before installing from another source; private GitHub sources can reuse local Git credentials for update checks
 - **Projects and agents** — View global and project Skills together, filter by agent, manage which agents can read a Skill, and copy Project Skills to other projects; use built-in agent definitions or add your own
 - **Cross-platform management** — Manage Skills on Windows, macOS, and Linux; Windows users can also switch to installed WSL distributions
 
@@ -102,7 +102,7 @@ Download the latest release from [GitHub Releases](https://github.com/hccake/ski
 
 ### 1. Choose an install entry point
 
-Choose a Skill from the Discover page, or open the install entry in Global Skills or a target project. When installing a Project Skill, add or select the project in the sidebar first.
+Choose a Skill from the Discover page to open the installation wizard in the current environment. The wizard defaults to Global Skills; you can select an added project to install there. You can also open the install entry directly from Global Skills or a project on the Skills page.
 
 ### 2. Provide a Skill source
 
@@ -121,7 +121,7 @@ npx skills add vercel-labs/agent-skills --skill frontend-design -a claude-code
 
 ### 3. Confirm and install
 
-Select the Skills, target agents, and installation mode, review the change preview, and run the installation. Cross-host download redirects require explicit confirmation. Direct downloads are one-time, new-install sources and do not provide update, reinstall, or source-repair actions afterward. After installation, the selected agents can read the Skill; you can continue reading and managing agents from the Skills workspace.
+Select the Skills, target agents, and installation mode, review the change preview, and run the installation. Cross-host download redirects require explicit confirmation. Direct downloads are one-time, new-install sources and do not provide updates; remove the installed Skill before installing it again. After installation, the selected agents can read the Skill; you can continue reading and managing agents from the Skills workspace.
 
 ---
 

README.zh-CN.md

@@ -37,7 +37,7 @@
 ## ✨ 核心能力
 
 - **发现与安装**：从在线目录发现 Skill，查看来源、说明和安全信息，或者从 GitHub、Git、本地目录、约定地址（Well-known 地址）、原始 `SKILL.md` 文件和 ZIP、tar 归档安装
-- **浏览与维护**：阅读已安装 Skill 的完整内容，检查并执行更新，在来源失效时重新选择来源；私有 GitHub 来源可以复用本机 Git 凭据检查更新
+- **浏览与维护**：阅读已安装 Skill 的完整内容，检查并执行更新；来源失效时可以保留当前 Skill，或移除后从其他来源重新安装；私有 GitHub 来源可以复用本机 Git 凭据检查更新
 - **项目与 Agent 管理**：统一查看全局与项目 Skill，按 Agent 筛选，调整 Skill 可供哪些 Agent 使用，并在项目之间复制 Skill；既可以使用随应用提供的 Agent 信息，也可以添加自己的 Agent 信息
 - **跨平台管理**：在 Windows、macOS 和 Linux 上管理 Skill；Windows 用户还可以切换到已安装的 WSL 发行版
 
@@ -101,7 +101,7 @@
 
 ### 1. 选择安装入口
 
-可以从“发现”页面选择 Skill，也可以在“全局 Skill”或目标项目中打开安装入口。安装项目 Skill 时，先在侧栏添加或选择对应项目。
+从“发现”页面选择 Skill 后，安装向导使用当前环境并默认选中全局，用户可以改选已添加的项目。也可以直接从 Skills 页的“全局 Skill”或项目区域打开安装入口。
 
 ### 2. 提供 Skill 来源
 
@@ -120,7 +120,7 @@ npx skills add vercel-labs/agent-skills --skill frontend-design -a claude-code
 
 ### 3. 确认并安装
 
-选择需要安装的 Skill、目标 Agent 和安装方式，检查变更预览，然后执行安装。下载地址发生跨主机重定向时，需要明确确认最终下载主机。直接下载属于一次性全新安装来源，安装后不提供更新、重新安装或来源修复操作。安装完成后，关联的 Agent 即可读取该 Skill；你也可以在 Skills 工作台中继续阅读和管理 Agent。
+选择需要安装的 Skill、目标 Agent 和安装方式，检查变更预览，然后执行安装。下载地址发生跨主机重定向时，需要明确确认最终下载主机。直接下载属于一次性全新安装来源，安装后不提供更新；再次安装前需要先移除现有 Skill。安装完成后，关联的 Agent 即可读取该 Skill；你也可以在 Skills 工作台中继续阅读和管理 Agent。
 
 ---
 

docs/README.md

@@ -10,7 +10,7 @@ Skill Deck 的长期文档说明当前产品、领域规则和系统约束。每
 | [产品行为与交互](./product.md) | 用户看到的核心概念、页面能力、用户工作流、反馈规则、Agent 筛选交互和产品限制 | 用户可见能力、交互语义或产品限制变化 |
 | [Agent 模型](./agent-model.md) | Agent 注册表、Skill 读取位置、Agent 检测位置、关联 Agent、筛选候选、选择分组和安装初始选择 | Agent 模型、解析规则或选择语义变化 |
 | [Environment、Skill 位置与项目管理](./environments-and-projects.md) | 应用所在系统、Windows 与 WSL 切换、全局 Skill、项目 Skill、已添加项目、路径解析和跨文件系统访问 | Windows/WSL 切换、全局或项目行为、已添加项目、路径或文件访问规则变化 |
-| [Skill 生命周期](./skill-lifecycle.md) | 来源、发现、安装、读取、更新、来源修复、管理 Agent、复制和移除 | Skill 从来源进入本地后的业务流程变化 |
+| [Skill 生命周期](./skill-lifecycle.md) | 来源、发现、安装、读取、更新、管理 Agent、复制和移除 | Skill 从来源进入本地后的业务流程变化 |
 | [更新检查](./update-checking.md) | 远端版本比较、缓存、重试、限流、凭据和状态文件 | 远端版本比较或检查策略变化 |
 | [系统架构](./architecture.md) | 系统边界、应用内部结构、进程与窗口、IPC 契约、平台适配、数据归属和安全保证 | 顶层模块、典型调用方向、窗口、传输接口、平台实现或数据归属变化 |
 | [测试与验证规范](./testing.md) | 测试类型、覆盖范围、测试夹具、跨平台规则和断言原则 | 测试架构、平台测试能力、测试夹具或验证策略变化 |
@@ -28,7 +28,7 @@ Skill Deck 的长期文档说明当前产品、领域规则和系统约束。每
 | 修改用户可见功能、页面交互或 Agent 筛选 | [产品行为与交互](./product.md)，再按涉及概念进入对应主文档 |
 | 修改 Skill 读取位置、Agent 检测位置、关联 Agent、筛选候选、选择分组或安装初始选择 | [Agent 模型](./agent-model.md) |
 | 修改应用所在系统、Windows/WSL 切换、全局或项目行为、已添加项目、路径解析或跨文件系统访问 | [Environment、Skill 位置与项目管理](./environments-and-projects.md) |
-| 修改来源、发现、安装、更新、来源修复、管理 Agent、复制或移除 | [Skill 生命周期](./skill-lifecycle.md)；涉及远端检查时增加[更新检查](./update-checking.md)，涉及写入机制时增加[执行与恢复](./execution-and-recovery.md) |
+| 修改来源、发现、安装、更新、管理 Agent、复制或移除 | [Skill 生命周期](./skill-lifecycle.md)；涉及远端检查时增加[更新检查](./update-checking.md)，涉及写入机制时增加[执行与恢复](./execution-and-recovery.md) |
 | 修改路由、跨窗口通信、Tauri 命令、WSL 传输或顶层依赖 | [系统架构](./architecture.md) |
 | 新增或修改测试、测试夹具、平台适配器验证或 E2E | [测试与验证规范](./testing.md)；修改 CI 命令时增加[贡献指南](../CONTRIBUTING.md) |
 | 跟踪 `skills` CLI 版本变化 | [skills CLI 参考与兼容](./skills-cli-reference.md)，再按变化类型进入受影响的领域主文档 |

docs/adr/0010-use-single-writer-library-persistence.md

@@ -0,0 +1,11 @@
+---
+status: accepted
+---
+
+# Skill Library 采用单写者持久化模型
+
+Skill Library catalog 和成员目录只由当前 Skill Deck 进程写入。Tauri 单实例机制限制正常运行时的应用实例数量，`RuntimeAdmissionCoordinator` 统一控制持久化写入许可，Runtime Library Adapter 再按 Environment 串行化 catalog、成员内容和内部恢复 I/O。其他进程、命令行工具和用户直接修改 catalog 不属于受支持的写入方式。
+
+这项取舍删除了 Native 操作系统文件锁和 WSL `flock` 守卫协议，同时保留成员条件提交、catalog 原子写入、目录备份和崩溃恢复。全局和项目 Skill 的兼容 lock 仍可能被 `skills` CLI 或用户工具修改，因此继续在提交前重新读取并检查修订。
+
+正式支持多个 Skill Deck 实例、增加独立后台写入进程、允许其他产品写入 Library，或者让多个设备共享同一 Library 数据目录时，必须重新讨论跨进程协调协议。当前行为由[系统架构](../architecture.md#应用内部结构)、[执行与恢复](../execution-and-recovery.md#原子写入与-lock-提交)和[测试与验证规范](../testing.md#http时间与外部进程)负责说明。