Real Skill packageSource verifiedClawHub registry

windows-shell-routing

Windows 上 AI agent 的 shell 选型规范：默认走 Git Bash，只在明确边界处单条命令切到 PowerShell 或 WSL。给出 12 类任务的 shell 归属、必须切 PowerShell 的四类、WSL 值不值得上的判定，以及 Git Bash 的五个坑及规避写法。适用于在 Windows 10/11 上做 JS/TS、Python 等开发的 agent 会话；选定 shell 之后的编码处理见 windows-shell skill。

Identity and source

Publisher attributionChenmoregistry owner unverified by skillvetai
Functional categoryAgent Engineering, Security & Governanceautomatically inferred · 60% rule confidence
Package forminstruction bundle3 recorded files
Canonical sourceClawHub registryclawhub:chenmo0414:windows-shell-routing
Open canonical source ↗

Platform declarations

These states come from the source or distribution context. None of the entries below are SkillVetAI compatibility test results.

OpenClawnative officialProvenance: registry distribution

Independent structural checks

These checks parse the fixed package against dated platform rules. They do not execute the Skill or verify task behavior.

Claude Codepasses structure
Checker 0.1.0 · agent-skills-2026-08-13+claude-code-docs-2026-08-13 · 8/19/2026.claude/skills/windows-shell-routing

Runtime, accounts, dependencies, permissions, network behavior and task quality remain untested.

OpenAI Codexpasses structure
Checker 0.1.0 · agent-skills-2026-08-13+codex-docs-2026-08-13 · 8/19/2026.agents/skills/windows-shell-routing

Runtime, accounts, dependencies, permissions, network behavior and task quality remain untested.

OpenClawissues found
Checker 0.1.0 · agent-skills-2026-08-13+openclaw-docs-2026-08-13 · 8/19/2026skills/windows-shell-routing
1 structural issue
  • error: metadata.openclaw.os must contain only darwin, linux, or win32. SKILL.md

Runtime, accounts, dependencies, permissions, network behavior and task quality remain untested.

Installation and inspection

This command is recorded from the source ecosystem and resolves the registry's latest release. The fixed release shown on this page should be inspected before adoption.

clawhub install @chenmo0414/windows-shell-routing
clawhub inspect @chenmo0414/windows-shell-routing --version 1.0.0

Security evidence

SkillVetAI static result: no findings detected

This automated, non-executing scan is bound to this release hash. It is not a safety certification and may contain false positives or false negatives.

Status
completed
Coverage
full text content
Files
3 / 3 inspected as text
Checked
8/19/2026, 9:53:13 AM
Scanner
0.1.3
Policy
1.0.3
3 inferred permission indicators
  • shell execution — automatically inferred
  • network access — automatically inferred
  • filesystem write — automatically inferred
1 dependency and API indicators
  • api: clawhub.ai
External clawhub result: clean

This is registry-supplied evidence for the recorded release, not an independent SkillVetAI scan. Check the canonical source for the full report, scanner versions, scope, and current moderation state.

Evidence checked
8/19/2026, 8:58:09 AM
Release binding
Matches this record
  • vt: clean
  • skillspector: clean
  • llm: clean

Recorded files

The catalog stores hashes and an inventory summary for change detection. It does not republish the package contents.

Package content hashsha256:baf2c80f193f83ed4fded3ada55228152c6bea99785b020ead0334ca3fda0fa8
Show up to 3 recorded paths
  • CHANGELOG.md
  • skill-card.md
  • SKILL.md

Source changelog

首个版本。Windows 上 AI agent 的 shell 选型规范：默认走 Git Bash，只在明确边界处 **单条命令**切到 PowerShell 或 WSL，主线不搬家。 与本工厂的 `windows-shell` 互补，不重叠——那个管「怎么正确执行命令」（编码）， 这个管「该用哪个 shell」（选型）。选定之后的编码处理直接引用前者，不复述细节。 两者 18 个技术主题里只有 3 个交叉，且细节都落在 `windows-shell` 一侧。 依据来自同一台 Windows 10 机器上的四臂 agent 对照实验——四个独立 agent 做同一份任务 （zod + TypeScript + vitest，pydantic + pytest，共 9 步），每臂只准用一种 shell： | 执行方式 | 输出 token | 轮次 | 失败 | 重试 | |------|------|------|------|------| | Git Bash | 8,114（基准） | 34 | 0 | 0 | | WSL → `/mnt/d` | 11,543（1.42x） | 34 | 0 | 1（静默） | | WSL → ext4 | 15,159（1.87x） | 49 | 1 | 2 | | pwsh 7 | 18,808（2.32x） | 46 | 0 | 0 | 另有 12 场景合成基准独立复现同一系数（PS 5.1 输出体积是 POSIX 的 2.22 倍）， 以及命令启动开销测量（Git Bash 0.064s / `wsl.exe` 0.130s / PS 5.1 0.189s / pwsh 7 0.216s）。 内容： - **决策表**：12 类常见任务的 shell 归属。含一条反直觉结论——native 模块编译 （node-gyp + MSVC）用 Git Bash 即可，与 pwsh **完全等价**：产物字节数相同， 都经 vswhere 找到同一套 VS 工具链。 - **必须切 PowerShell 的四类**：系统层面、…