Updates documentation for Windows security features

README.md:
- Add skip parameters example (-SkipVNC, -SkipOllama)
- Document VNC password prompt and minimum length
- Update requirements to show ISO creation fallbacks

CLAUDE.md:
- Add Windows script editing section
- Add Windows security patterns section
- Add Windows testing instructions
- Update VNC password minimum from 6 to 8 chars
- Document checksum verification for Windows

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>

CHANGELOG.md

@@ -0,0 +1,169 @@
+# Changelog
+
+All notable changes to this project will be documented in this file.
+
+The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.1.0/),
+and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
+
+## [0.14.1] - 2025-01-25
+
+### Changed
+- Updated README.md with Windows skip parameters and VNC password info
+- Updated CLAUDE.md with Windows script documentation and security patterns
+- Fixed VNC password minimum documentation (6 → 8 chars)
+
+## [0.14.0] - 2025-01-25
+
+### Security
+- Add SHA256 checksum verification for Ubuntu cloud image downloads (prevents MITM attacks)
+- Add strict input validation for git name, email, VM password, and VNC password
+- Validate loaded config.env values to detect tampering
+- VNC password minimum increased from 6 to 8 characters
+- Block shell metacharacters in all user inputs to prevent injection
+- Config file created with restricted ACL from the start (no race condition window)
+- Add security reminder to remove cloud-init ISO after first boot (contains passwords)
+
+### Added
+- ARM64 architecture detection for Windows on ARM devices
+- Log file creation at `$env:TEMP\setup_env_windows_<timestamp>.log`
+- Cleanup on failure: automatically removes partial VM, disk, and ISO on error
+- Hosts file backup before modification (removed on success, kept on failure)
+- Input validation functions: `Test-GitName`, `Test-GitEmail`, `Test-VMPassword`, `Test-VNCPassword`
+- Checksum caching to avoid re-downloading verification data
+
+### Changed
+- Ubuntu image URL now uses detected architecture instead of hardcoded amd64
+- All major operations now log to file for troubleshooting
+- VM creation wrapped in try/catch with automatic cleanup on failure
+
+## [0.13.0] - 2025-01-25
+
+### Changed
+- Rewrote `setup_env_windows.ps1` to fully implement WINDOWS_PLAN.md
+- Password handling uses cloud-init `chpasswd` with plaintext (type: text) instead of broken hash generation
+- Multiple ISO creation methods with fallback chain: oscdimg → WSL genisoimage → IMAPI2 COM
+- Downloads use BITS transfer for reliability with progress reporting
+- SSH readiness checking with timeout before displaying connection info
+
+### Added
+- Component skip parameters: `-SkipVNC`, `-SkipPostgreSQL`, `-SkipOllama`, `-SkipPlaywright`
+- VNC password support via base64 encoding in cloud-init
+- Automatic hosts file cleanup when VM is deleted with `-Force`
+- Proper prerequisite checking for Hyper-V, Windows edition, and admin privileges
+
+### Fixed
+- Cloud-init password configuration (was using bash syntax in PowerShell)
+- ISO creation now works without Windows ADK by using WSL or IMAPI2 fallbacks
+- Hosts file handling with proper admin privilege elevation
+
+## [0.12.0] - 2025-01-25
+
+### Added
+- Python runtime management via mise (alongside Node.js, Erlang, Elixir)
+- `WINDOWS_PLAN.md` documenting Hyper-V implementation strategy and security rationale
+
+### Fixed
+- Tidewave CLI download URL (now uses correct `tidewave_app` repo with musl binaries)
+
+### Changed
+- Python is now a selectable component managed by mise instead of system apt
+
+## [0.11.0] - 2025-01-25
+
+### Added
+- Windows support via Hyper-V for maximum security isolation
+- `setup_env_windows.ps1` PowerShell script with full VM provisioning
+- Ubuntu cloud image support with cloud-init automation
+- SSH key generation for passwordless VM access on Windows
+- Hosts file integration for easy `<vmname>.local` access
+
+### Security
+- Hyper-V provides stronger isolation than WSL2 (separate kernel, network, filesystem)
+- No host integration services enabled by default
+
+## [0.10.0] - 2025-01-25
+
+### Added
+- OpenCode: Open-source AI coding assistant with multi-provider support
+- Tidewave CLI: Elixir/Phoenix MCP server for AI-powered development
+- New component selection options for OpenCode and Tidewave
+
+## [0.9.1] - 2025-01-25
+
+### Fixed
+- Add error checking for base64 decode operations in VM provisioning
+- Add `set -e` to VM bootstrap script for early failure detection
+
+## [0.9.0] - 2025-01-25
+
+### Security
+- Fix rustup pipe-to-shell vulnerability: now downloads to temp file with validation before execution
+- Fix SKIP_EXPORTS command injection risk: refactored to use base64-encoded list instead of shell command string
+- Fix Playwright symlink path validation: validates executable and path prefix before creating symlinks
+
+## [0.8.0] - 2025-01-25
+
+### Added
+- Project memory system using CLAUDE.md, CHANGELOG.md, and README.md
+- Versioning rules and documentation update guidelines in CLAUDE.md
+
+## [0.7.0] - 2025-01-25
+
+### Added
+- CHANGELOG.md with version history following Keep a Changelog format
+
+## [0.6.0] - 2025-01-25
+
+### Changed
+- All tools now use latest versions by default instead of pinning specific versions
+- PostgreSQL authentication uses scram-sha-256 for all TCP connections
+- Simplified tool installation by removing version pinning constraints
+
+### Security
+- VNC passwords are never stored and must be entered each time
+- Added documentation for input validation patterns and safe config loading
+
+## [0.5.0] - 2025-01-25
+
+### Security
+- Prevents shell injection through input validation and safe parameter passing
+- Replaces direct sourcing with manual config parsing to avoid code execution
+- Downloads and validates install scripts before execution instead of piping
+- Uses base64 encoding for secure VM parameter transmission
+- Adds checksum verification for binary downloads
+- Creates secure temporary directories and files with proper permissions
+
+## [0.4.0] - 2025-01-25
+
+### Changed
+- Replaces sequential installation with parallel step execution
+- Introduces real-time progress dashboard with spinner and status
+- Removes color variables to improve terminal compatibility
+- Restructures logging with per-step files for better debugging
+
+### Performance
+- Significantly reduces total setup time by running independent steps concurrently
+
+## [0.3.0] - 2025-01-25
+
+### Added
+- Dual-mode operation: orchestration on macOS, provisioning on Linux
+- Interactive component selection with visual menu interface
+- VNC desktop access for OAuth workflows and browser-based tasks
+
+### Security
+- Secure VM creation with disabled host filesystem access
+
+## [0.2.0] - 2025-01-25
+
+### Added
+- OrbStack development sandbox setup script
+- mise version manager with Node.js, Erlang, and Elixir support
+- PostgreSQL 16 with remote access configuration
+- Claude Code integration with multiple plugin marketplaces
+- Chromium browser and Playwright for automation tasks
+
+## [0.1.0] - 2025-01-25
+
+### Added
+- Initial project structure

CLAUDE.md

@@ -1,17 +1,42 @@
 # CLAUDE.md
 
+## Project Memory
+
+This project uses three documentation files as persistent memory. **You must keep these files up to date when making changes:**
+
+| File | Purpose | Update When |
+|------|---------|-------------|
+| `CLAUDE.md` | Technical context for Claude Code sessions | Adding new patterns, conventions, or implementation details |
+| `CHANGELOG.md` | Version history following Keep a Changelog format | Every commit (add entry, bump version) |
+| `README.md` | User-facing documentation | Changing user-visible behavior, adding features, or modifying usage |
+
+### Versioning Rules
+
+- Use semantic versioning (MAJOR.MINOR.PATCH)
+- Increment PATCH for bug fixes and minor updates
+- Increment MINOR for new features and enhancements
+- Increment MAJOR for breaking changes
+- Tag every commit with its version: `git tag -a vX.Y.Z -m "Description"`
+
 ## Project Overview
 
-This repository contains a dual-mode setup script for creating OrbStack-based development sandboxes tailored for Claude Code with Elixir/Erlang, browser automation, and PostgreSQL.
+This repository provides scripts for creating isolated development sandboxes for Claude Code on macOS (OrbStack) and Windows (Hyper-V). Both platforms offer full VM isolation for safely running AI coding assistants with elevated permissions.
+
+**Supported platforms:**
+- **macOS**: OrbStack VMs (ARM64 Apple Silicon)
+- **Windows**: Hyper-V VMs (maximum security, stronger than WSL2)
 
 ## Repository Structure
 
 ```
-setup_env.sh        - Main script (macOS host mode + Linux VM provisioning mode)
+setup_env.sh            - macOS script (OrbStack host mode + Linux VM provisioning)
+setup_env_windows.ps1   - Windows script (Hyper-V with cloud-init)
+WINDOWS_PLAN.md         - Windows implementation plan and security rationale
 config.env.example      - Example credentials file
 config.env              - User credentials (gitignored, created on first run)
 .gitignore              - Ignores config.env
 README.md               - User-facing documentation
+CHANGELOG.md            - Version history (Keep a Changelog format)
 CLAUDE.md               - This file (context for Claude Code sessions)
 ```
 
@@ -61,7 +86,7 @@ This means `./setup_env.sh my-vm` on macOS does everything end-to-end.
 - `--yes`/`-y` accepts all components without prompting but still allows interactive credential entry
 - `MISE_GLOBAL_CONFIG_FILE` and `MISE_CONFIG_DIR` are set to prevent OrbStack host-mount config pollution
 
-### Security Patterns
+### Security Patterns (macOS/Linux)
 
 - **Input validation**: Use `validate_vm_name()`, `validate_vnc_password()`, `validate_safe_input()` for all user inputs
 - **Safe config loading**: Use `load_config_safely()` which parses key=value pairs without shell execution (never `source`)
@@ -71,9 +96,30 @@ This means `./setup_env.sh my-vm` on macOS does everything end-to-end.
 - **Symlinks**: Check with `[ ! -e path ]` (not `[ ! -L path ]`) to avoid overwriting existing files
 - **Architecture detection**: Use `detect_architecture()` for portable binary downloads
 
+### Editing setup_env_windows.ps1
+
+- PowerShell script for Windows Hyper-V VM creation
+- Uses Ubuntu cloud images with cloud-init for automated provisioning
+- Supports ARM64 architecture detection for Windows on ARM
+- Skip parameters: `-SkipVNC`, `-SkipPostgreSQL`, `-SkipOllama`, `-SkipPlaywright`
+- ISO creation fallback chain: oscdimg → WSL genisoimage → IMAPI2 COM
+- Log file created at `$env:TEMP\setup_env_windows_<timestamp>.log`
+
+### Security Patterns (Windows)
+
+- **Input validation**: Use `Test-GitName`, `Test-GitEmail`, `Test-VMPassword`, `Test-VNCPassword` for all user inputs
+- **Config file security**: ACL set to owner-only before writing content (no race condition)
+- **Config validation**: Loaded config.env values are re-validated to detect tampering
+- **Checksum verification**: Ubuntu image verified against SHA256 checksums from Canonical
+- **Cleanup on failure**: `Invoke-Cleanup` removes partial VM, disk, and ISO on errors
+- **Hosts file backup**: Created before modification, removed on success, kept on failure
+- **Cloud-init ISO**: Contains passwords in plaintext; user reminded to remove after first boot
+
 ### Testing
 
 There are no automated tests. To test changes:
+
+**macOS:**
 1. Remove an existing test VM: `orb delete test-sandbox`
 2. Run: `./setup_env.sh test-sandbox`
 3. Verify provisioning completes
@@ -81,13 +127,21 @@ There are no automated tests. To test changes:
 5. Test VNC: `ssh test-sandbox@orb -- vnc-start`, then `open vnc://test-sandbox.orb.local:5901`
 6. Clean up: `orb delete test-sandbox`
 
+**Windows:**
+1. Remove existing test VM: `Stop-VM -Name test-sandbox -Force; Remove-VM -Name test-sandbox -Force`
+2. Run: `.\setup_env_windows.ps1 -VMName test-sandbox`
+3. Verify provisioning completes (check `/var/log/provision.log` in VM)
+4. Test SSH: `ssh -i $env:USERPROFILE\.ssh\id_ed25519_test-sandbox dev@test-sandbox.local`
+5. Clean up: `Stop-VM -Name test-sandbox -Force; Remove-VM -Name test-sandbox -Force`
+
 ### Security Considerations
 
-- `config.env` is chmod 600 and gitignored (stores git name/email only, never VNC password)
+- `config.env` is chmod 600 / owner-only ACL and gitignored (stores git name/email only, never passwords)
 - PostgreSQL uses scram-sha-256 for all TCP connections (peer auth for local socket)
 - The script refuses to run as root inside the VM
-- VNC password is required (min 6 chars), validated to block shell metacharacters, never stored
+- VNC password is required (min 8 chars), validated to block shell metacharacters, never stored
-- VNC binds to all interfaces (`-localhost no`) to allow connections from the macOS host — this is intentional for the OrbStack use case and documented in the vnc-start script
+- VNC binds to all interfaces (`-localhost no`) to allow connections from the host — this is intentional and documented
 - All user inputs are validated before use to prevent command injection
 - External scripts (mise, ollama) are downloaded to temp files and validated before execution
 - Host filesystem access is disabled inside the VM for isolation
+- Windows: Ubuntu image downloads are verified against SHA256 checksums

README.md

@@ -1,20 +1,27 @@
-# OrbStack Development Sandbox
+# Secure AI Coding Sandboxes
 
-sDisposable, isolated Linux VMs for running Claude Code with `--dangerously-skip-permissions`. One command creates a fully provisioned environment. Blow it away and recreate it in minutes.
+Disposable, isolated Linux VMs for running Claude Code with `--dangerously-skip-permissions`. One command creates a fully provisioned environment. Blow it away and recreate it in minutes.
 
-The VM is a real Linux machine with its own filesystem, network, and process space — but you edit files from your Mac, access services on `*.orb.local`, and SSH in without any key setup. All the isolation of a container with none of the friction.
+The VM is a real Linux machine with its own filesystem, network, and process space — complete isolation from your host with easy access for development.
+
+| Platform | Script | Isolation Level |
+|----------|--------|-----------------|
+| **macOS** | `setup_env.sh` (OrbStack) | Full VM isolation |
+| **Windows** | `setup_env_windows.ps1` (Hyper-V) | Full VM isolation |
 
 ## Why This Exists
 
 Running `claude --dangerously-skip-permissions` on your host machine means Claude can execute arbitrary commands, install packages, modify system files, and access everything on your disk. That's powerful for autonomous coding but risky on a machine with your SSH keys, credentials, and personal files.
 
-This script creates throwaway VMs where Claude can run unrestricted:
+These scripts create throwaway VMs where Claude can run unrestricted:
 
-- **Isolated filesystem** — Claude can't touch your macOS files, keys, or configs
+- **Isolated filesystem** — Claude can't touch your host files, keys, or configs
 - **Isolated network** — services run on their own IP, no port conflicts with your host
-- **Disposable** — `orb delete my-sandbox` wipes everything; recreate in one command
+- **Disposable** — delete and recreate in one command
 - **Multiple VMs** — run separate sandboxes per project with shared git credentials
-- **Full access from Mac** — edit files in your editor, browse databases, view running apps
+- **Full access from host** — edit files in your editor, browse databases, view running apps
+
+## Quick Start (macOS)
 
 ```bash
 # Create a sandbox, SSH in, run Claude unrestricted
@@ -25,7 +32,7 @@ claude --dangerously-skip-permissions
 
 If anything goes wrong: `orb delete my-project && ./setup_env.sh my-project`
 
-## Quick Start
+### macOS Setup
 
 ```bash
 # Create and provision a VM (one command from macOS)
@@ -51,11 +58,65 @@ When run manually inside a VM, you're prompted for each component individually:
 ./setup_env.sh --yes
 ```
 
-## Requirements
+### macOS Requirements
 
 - macOS with Apple Silicon (ARM64)
 - [OrbStack](https://orbstack.dev) installed (`brew install orbstack`)
 
+## Quick Start (Windows)
+
+```powershell
+# Run as Administrator
+.\setup_env_windows.ps1 -VMName my-project
+
+# Connect via SSH (after provisioning)
+ssh -i $env:USERPROFILE\.ssh\id_ed25519_my-project dev@my-project.local
+
+# Run Claude unrestricted
+claude --dangerously-skip-permissions
+```
+
+If anything goes wrong: `Remove-VM -Name my-project -Force` then run the script again.
+
+### Windows Setup
+
+```powershell
+# Create and provision a VM (run as Administrator)
+.\setup_env_windows.ps1 -VMName my-sandbox
+
+# Customize resources
+.\setup_env_windows.ps1 -VMName my-sandbox -MemoryGB 16 -DiskGB 100 -CPUs 8
+
+# Skip optional components
+.\setup_env_windows.ps1 -VMName my-sandbox -SkipVNC -SkipOllama
+```
+
+On first run, you'll be prompted for:
+- Git commit author name and email (saved to `config.env`)
+- VM user password (min 8 chars, not stored)
+- VNC password if VNC is enabled (min 8 chars, not stored)
+
+### Windows Requirements
+
+- Windows 10/11 Pro, Enterprise, or Education (Hyper-V not available on Home)
+- Hyper-V enabled: `Enable-WindowsOptionalFeature -Online -FeatureName Microsoft-Hyper-V -All`
+- Administrator privileges
+- One of: Windows ADK, WSL with genisoimage, or Windows 10+ (IMAPI2 fallback)
+
+### Why Hyper-V (not WSL2)?
+
+WSL2 is convenient but provides weaker isolation:
+- Shares kernel with all WSL2 instances
+- Host filesystem mounted by default
+- Can launch Windows executables from Linux
+- Network traffic bypasses Windows firewall
+
+Hyper-V provides **maximum security**:
+- Separate kernel per VM
+- Complete filesystem isolation
+- Own network stack
+- No Windows integration by default
+
 ## What Gets Installed
 
 All components are optional — deselect what you don't need in the interactive picker.
@@ -70,7 +131,9 @@ All components are optional — deselect what you don't need in the interactive
 | Playwright | latest | Browser testing framework |
 | PostgreSQL | system default | Database |
 | Ollama | latest | Local LLM inference |
-| Claude Code | latest | AI coding assistant |
+| Claude Code | latest | AI coding assistant (Anthropic) |
+| OpenCode | latest | Open-source AI coding assistant (multi-provider) |
+| Tidewave | latest | Elixir/Phoenix MCP server for AI tools |
 | yq | latest | YAML processor |
 | watchexec | latest | File watcher (via cargo) |
 | TigerVNC + XFCE | system | VNC access for browser login flows |

WINDOWS_PLAN.md

@@ -0,0 +1,257 @@
+# Windows Hyper-V Implementation Plan
+
+This document describes the plan for implementing secure, isolated development sandboxes on Windows using Hyper-V instead of WSL2.
+
+## Why Hyper-V Over WSL2
+
+### WSL2 Security Limitations
+
+| Issue | Impact |
+|-------|--------|
+| **Shared kernel** | All WSL2 instances share Microsoft's Linux kernel |
+| **Host filesystem access** | `/mnt/c` mounted by default, can access Windows files |
+| **Bidirectional execution** | Linux can launch Windows executables |
+| **Firewall bypass** | WSL2 outbound traffic bypasses Windows firewall rules |
+| **Process visibility** | Windows has limited visibility into WSL2 processes |
+| **Credential exposure** | Linux processes can potentially access Windows credentials |
+
+### CVE History
+
+- **CVE-2024-20681**: Local privilege escalation to SYSTEM via WSL
+- **CVE-2025-9074**: Container escape via WSL2 filesystem sharing
+- **CVE-2025-53788**: Undisclosed flaw in WSL/host communication
+
+### Hyper-V Isolation Model
+
+| Feature | Benefit |
+|---------|---------|
+| **Separate kernel** | Each VM runs its own Linux kernel |
+| **Complete filesystem isolation** | No access to Windows files by default |
+| **Own network stack** | Separate IP, no firewall bypass |
+| **No Windows integration** | Cannot launch Windows programs |
+| **Snapshot support** | Can checkpoint and restore VM state |
+| **Hardware isolation** | Uses VT-x/AMD-V virtualization |
+
+## Implementation Architecture
+
+```
+┌─────────────────────────────────────────────────────────────────┐
+│                      Windows Host                                │
+│                                                                  │
+│  ┌─────────────────────┐    ┌─────────────────────────────────┐ │
+│  │ setup_env_windows.ps1│    │        Hyper-V Manager          │ │
+│  │                     │    │                                  │ │
+│  │ • Check prereqs     │───▶│  ┌──────────────────────────┐  │ │
+│  │ • Download image    │    │  │      Ubuntu VM           │  │ │
+│  │ • Create cloud-init │    │  │                          │  │ │
+│  │ • Provision VM      │    │  │  • Isolated filesystem   │  │ │
+│  │ • Configure network │    │  │  • Own network (NAT)     │  │ │
+│  └─────────────────────┘    │  │  • Claude Code + tools   │  │ │
+│                              │  │  • No Windows access     │  │ │
+│                              │  └──────────────────────────┘  │ │
+│                              └─────────────────────────────────┘ │
+└─────────────────────────────────────────────────────────────────┘
+```
+
+## Script Workflow
+
+### Phase 1: Prerequisites Check
+
+```powershell
+# Verify environment
+1. Check Administrator privileges
+2. Check Windows edition (Pro/Enterprise/Education required)
+3. Check Hyper-V enabled
+4. Check for existing VM (handle -Force flag)
+5. Find suitable network switch
+```
+
+### Phase 2: Configuration
+
+```powershell
+# Collect credentials
+1. Load config.env if exists (git name/email)
+2. Prompt for missing values
+3. Prompt for VM password (not stored)
+4. Generate SSH key pair for VM access
+```
+
+### Phase 3: Image Preparation
+
+```powershell
+# Prepare Ubuntu cloud image
+1. Download Ubuntu cloud image (VHDX format)
+2. Cache in $env:ProgramData\HyperV-DevSandbox\Images
+3. Copy and resize for new VM
+4. Generate cloud-init ISO with:
+   - User account configuration
+   - SSH key injection
+   - Package installation
+   - Setup script embedding
+```
+
+### Phase 4: VM Creation
+
+```powershell
+# Create Hyper-V VM
+1. Create Generation 2 VM (UEFI)
+2. Configure resources (CPU, RAM, disk)
+3. Disable Secure Boot (for Ubuntu)
+4. Attach cloud-init ISO
+5. Disable integration services for isolation
+6. Enable nested virtualization (for Docker)
+```
+
+### Phase 5: Provisioning
+
+```powershell
+# Start and provision
+1. Start VM
+2. Wait for IP assignment
+3. Add to hosts file (<vmname>.local)
+4. Cloud-init runs setup_env.sh internally
+5. Display connection instructions
+```
+
+## Cloud-Init Configuration
+
+```yaml
+#cloud-config
+hostname: <vmname>
+users:
+  - name: dev
+    sudo: ALL=(ALL) NOPASSWD:ALL
+    ssh_authorized_keys:
+      - <generated-public-key>
+packages:
+  - openssh-server
+  - curl
+  - git
+write_files:
+  - path: /tmp/setup_env.sh
+    content: <base64-encoded-script>
+runcmd:
+  - bash /tmp/setup_env.sh --non-interactive --yes
+```
+
+## Security Hardening
+
+### VM Configuration
+
+```powershell
+# Disable integration services for maximum isolation
+Disable-VMIntegrationService -Name "Guest Service Interface"
+Disable-VMIntegrationService -Name "Heartbeat"
+# Keep Time Synchronization and Shutdown for usability
+```
+
+### Network Isolation Options
+
+1. **NAT (Default)**: VM can access internet, isolated from LAN
+2. **Internal Only**: VM can only communicate with host
+3. **Private**: Completely isolated network
+
+### Firewall Rules (Optional)
+
+```powershell
+# Allow SSH access only
+New-NetFirewallRule -Name "HyperV-SSH-$VMName" `
+    -Direction Inbound -LocalPort 22 -Protocol TCP `
+    -RemoteAddress <VM-IP> -Action Allow
+```
+
+## Requirements
+
+### Windows Edition
+
+- Windows 10/11 Pro
+- Windows 10/11 Enterprise
+- Windows 10/11 Education
+- Windows Server 2016+
+
+**Not supported**: Windows 10/11 Home (no Hyper-V)
+
+### Hardware
+
+- 64-bit processor with SLAT (Second Level Address Translation)
+- VM Monitor Mode extensions (VT-x on Intel, AMD-V on AMD)
+- Minimum 4 GB RAM (8+ GB recommended)
+- BIOS/UEFI virtualization enabled
+
+### Software
+
+- Hyper-V enabled: `Enable-WindowsOptionalFeature -Online -FeatureName Microsoft-Hyper-V -All`
+- Windows ADK (for cloud-init ISO creation): [Download](https://docs.microsoft.com/en-us/windows-hardware/get-started/adk-install)
+
+## Usage
+
+### Create Sandbox
+
+```powershell
+# Basic usage (run as Administrator)
+.\setup_env_windows.ps1 -VMName my-project
+
+# With custom resources
+.\setup_env_windows.ps1 -VMName my-project -MemoryGB 16 -DiskGB 100 -CPUs 8
+
+# Replace existing VM
+.\setup_env_windows.ps1 -VMName my-project -Force
+```
+
+### Connect to Sandbox
+
+```powershell
+# Via SSH (recommended)
+ssh -i $env:USERPROFILE\.ssh\id_ed25519_<vmname> dev@<vmname>.local
+
+# Via Hyper-V console
+vmconnect localhost <vmname>
+```
+
+### Manage Sandbox
+
+```powershell
+# Stop VM
+Stop-VM -Name <vmname>
+
+# Start VM
+Start-VM -Name <vmname>
+
+# Get VM IP
+(Get-VM -Name <vmname> | Get-VMNetworkAdapter).IPAddresses
+
+# Delete VM completely
+Stop-VM -Name <vmname> -Force
+Remove-VM -Name <vmname> -Force
+Remove-Item -Path "$env:ProgramData\HyperV-DevSandbox\VMs\<vmname>" -Recurse -Force
+```
+
+## Comparison: OrbStack vs Hyper-V
+
+| Feature | OrbStack (macOS) | Hyper-V (Windows) |
+|---------|------------------|-------------------|
+| Host OS | macOS only | Windows only |
+| Setup complexity | Low | Medium |
+| Isolation | Full VM | Full VM |
+| DNS | `*.orb.local` auto | Manual hosts file |
+| SSH | Automatic | Key-based |
+| Filesystem sharing | Configurable | Disabled by default |
+| Performance | Near-native | Near-native |
+| Nested virtualization | Yes | Yes |
+
+## Known Limitations
+
+1. **Windows ADK Required**: Cloud-init ISO creation requires `oscdimg` from Windows ADK
+2. **Manual DNS**: No automatic DNS like OrbStack's `*.orb.local`
+3. **Key Management**: SSH keys generated per-VM, stored in user's .ssh folder
+4. **No Clipboard Sharing**: Disabled for security (can be enabled if needed)
+5. **Generation 2 Only**: Uses UEFI, no legacy BIOS support
+
+## Future Improvements
+
+- [ ] Add option for WSL2 (for users who prefer convenience over security)
+- [ ] Implement automatic DNS via Windows hosts file or local DNS server
+- [ ] Add PowerShell ISO creation without Windows ADK dependency
+- [ ] Support for VM templates/checkpoints
+- [ ] Integration with Windows Terminal for easy access
+- [ ] Optional WinRM configuration for PowerShell remoting

setup_env.sh

@@ -284,30 +284,34 @@ if [[ "$(uname -s)" == "Darwin" ]]; then
 
     # ---- Interactive component selection ----
     # Component IDs and descriptions
-    COMP_IDS=(postgresql mise node erlang chromium vnc ollama claude playwright plugins)
+    COMP_IDS=(postgresql mise node python erlang chromium vnc ollama claude opencode playwright plugins)
     COMP_NAMES=(
         "PostgreSQL"
         "mise"
         "Node.js (LTS)"
+        "Python (latest)"
         "Erlang/Elixir"
         "Chromium"
         "VNC + XFCE"
         "Ollama"
         "Claude Code"
+        "OpenCode"
         "Playwright"
-        "Claude Plugins"
+        "Claude Plugins + Tidewave"
     )
     COMP_DESCS=(
         "Database server for local development"
-        "Version manager for Node.js, Erlang, Elixir"
+        "Version manager for Node.js, Python, Erlang, Elixir"
-        "JavaScript runtime (required for Claude Code)"
+        "JavaScript runtime (required for Claude/OpenCode)"
+        "Python runtime managed by mise"
         "BEAM VM + Elixir functional language"
         "Browser for automation and testing"
         "Remote desktop for browser-based login flows"
         "Local LLM runner for offline inference"
         "AI coding assistant CLI from Anthropic"
+        "Open-source AI coding assistant (multi-provider)"
         "Browser testing and automation framework"
-        "Code review, feature dev, browser automation plugins"
+        "Code review, Tidewave MCP, browser automation"
     )
 
     # All selected by default
@@ -390,12 +394,12 @@ if [[ "$(uname -s)" == "Darwin" ]]; then
 
     echo ""
 
-    # Build SKIP env var exports for unselected components
+    # Build list of skipped component IDs (space-separated, safe characters only)
-    SKIP_EXPORTS=""
+    # Security: Only hardcoded component IDs from COMP_IDS are used, no user input
+    SKIP_LIST=""
     for ((i=0; i<COMP_COUNT; i++)); do
         if [ "${COMP_SELECTED[$i]}" = "0" ]; then
-            UPPER_ID=$(echo "${COMP_IDS[$i]}" | tr '[:lower:]' '[:upper:]')
+            SKIP_LIST="${SKIP_LIST}${COMP_IDS[$i]} "
-            SKIP_EXPORTS="${SKIP_EXPORTS}export SKIP_${UPPER_ID}=1; "
         fi
     done
 
@@ -463,19 +467,32 @@ if [[ "$(uname -s)" == "Darwin" ]]; then
     orb push -m "$VM_NAME" "$SCRIPT_DIR/setup_env.sh" /tmp/setup_env.sh
 
     echo ""
-    if [ -n "$SKIP_EXPORTS" ]; then
+    if [ -n "$SKIP_LIST" ]; then
-        echo -e "${YELLOW}Skipping:${NC} $SKIP_EXPORTS"
+        echo -e "${YELLOW}Skipping:${NC} $SKIP_LIST"
     fi
     echo -e "${YELLOW}Running provisioning inside VM (this will take a while)...${NC}"
     echo ""
 
-    # Security: Use base64 encoding to safely pass values to VM (Issue #1)
+    # Security: Use base64 encoding to safely pass all values to VM
     # This prevents any shell injection through special characters
     GIT_NAME_B64=$(printf '%s' "$GIT_NAME" | base64)
     GIT_EMAIL_B64=$(printf '%s' "$GIT_EMAIL" | base64)
     VNC_PASSWORD_B64=$(printf '%s' "$VNC_PASSWORD" | base64)
+    SKIP_LIST_B64=$(printf '%s' "$SKIP_LIST" | base64)
 
-    orb run -m "$VM_NAME" bash -c "${SKIP_EXPORTS}export GIT_NAME=\$(echo '$GIT_NAME_B64' | base64 -d); export GIT_EMAIL=\$(echo '$GIT_EMAIL_B64' | base64 -d); export VNC_PASSWORD=\$(echo '$VNC_PASSWORD_B64' | base64 -d); bash /tmp/setup_env.sh --non-interactive"
+    # Security: All user-controlled values are base64-encoded before passing to VM
+    # The decode script sets SKIP_* env vars from the safe SKIP_LIST
+    orb run -m "$VM_NAME" bash -c "
+        set -e
+        export GIT_NAME=\$(echo '$GIT_NAME_B64' | base64 -d) || { echo 'Failed to decode GIT_NAME'; exit 1; }
+        export GIT_EMAIL=\$(echo '$GIT_EMAIL_B64' | base64 -d) || { echo 'Failed to decode GIT_EMAIL'; exit 1; }
+        export VNC_PASSWORD=\$(echo '$VNC_PASSWORD_B64' | base64 -d) || { echo 'Failed to decode VNC_PASSWORD'; exit 1; }
+        for comp in \$(echo '$SKIP_LIST_B64' | base64 -d); do
+            upper=\$(echo \"\$comp\" | tr '[:lower:]' '[:upper:]')
+            export \"SKIP_\${upper}=1\"
+        done
+        bash /tmp/setup_env.sh --non-interactive
+    "
 
     echo ""
     echo -e "${GREEN}============================================================================${NC}"
@@ -916,6 +933,15 @@ install_node() {
     export PATH="$HOME/.local/share/mise/shims:$PATH"
 }
 
+install_python() {
+    export PATH="$HOME/.local/bin:$PATH"
+    eval "$(~/.local/bin/mise activate bash)"
+    mise use -g python@latest
+    export PATH="$HOME/.local/share/mise/shims:$PATH"
+    # Ensure pip is available via mise python
+    python -m pip install --upgrade pip
+}
+
 install_erlang() {
     export PATH="$HOME/.local/bin:$PATH"
     eval "$(~/.local/bin/mise activate bash)"
@@ -1050,8 +1076,13 @@ install_playwright() {
 
     # Symlink Playwright's Chromium if no system one (check for existing files)
     if ! command_exists chromium-browser && ! command_exists chromium; then
-        local pw_chrome=$(find "$HOME/.cache/ms-playwright" -name "chrome" -type f 2>/dev/null | head -1)
+        local pw_chrome
-        if [ -n "$pw_chrome" ]; then
+        pw_chrome=$(find "$HOME/.cache/ms-playwright" -name "chrome" -type f 2>/dev/null | head -1)
+        # Security: Validate the found path before creating symlinks
+        # - Must not be empty
+        # - Must be an executable file
+        # - Must be within the expected playwright cache directory
+        if [ -n "$pw_chrome" ] && [ -x "$pw_chrome" ] && [[ "$pw_chrome" == "$HOME/.cache/ms-playwright"* ]]; then
             [ ! -e /usr/local/bin/chromium ] && sudo ln -sf "$pw_chrome" /usr/local/bin/chromium
             [ ! -e /usr/local/bin/google-chrome ] && sudo ln -sf "$pw_chrome" /usr/local/bin/google-chrome
             echo "Using Playwright's bundled Chromium"
@@ -1059,6 +1090,59 @@ install_playwright() {
     fi
 }
 
+install_opencode() {
+    export PATH="$HOME/.local/share/mise/shims:$HOME/.local/bin:$PATH"
+    if command_exists opencode; then
+        echo "OpenCode already installed"
+        return 0
+    fi
+    npm install -g opencode-ai
+}
+
+install_tidewave() {
+    if command_exists tidewave; then
+        echo "Tidewave already installed"
+        return 0
+    fi
+
+    # Download Tidewave CLI binary from tidewave_app releases
+    # Binary naming: tidewave-cli-<arch>-unknown-linux-musl
+    local machine_arch
+    machine_arch=$(uname -m)
+    local tidewave_arch=""
+    case "$machine_arch" in
+        x86_64)  tidewave_arch="x86_64" ;;
+        aarch64) tidewave_arch="aarch64" ;;
+        *)
+            log_error "Unsupported architecture for Tidewave: $machine_arch"
+            return 1
+            ;;
+    esac
+
+    # Get latest release URL from GitHub (tidewave_app repo)
+    local release_url="https://github.com/tidewave-ai/tidewave_app/releases/latest/download/tidewave-cli-${tidewave_arch}-unknown-linux-musl"
+
+    # Security: Download to temp file first, validate, then install
+    local temp_file
+    temp_file=$(mktemp)
+    if curl -fsSL "$release_url" -o "$temp_file"; then
+        # Verify it's an executable (ELF binary)
+        if file "$temp_file" | grep -q "ELF"; then
+            sudo mv "$temp_file" /usr/local/bin/tidewave
+            sudo chmod +x /usr/local/bin/tidewave
+            echo "Tidewave CLI installed"
+        else
+            log_error "Downloaded tidewave is not a valid binary"
+            rm -f "$temp_file"
+            return 1
+        fi
+    else
+        log_error "Failed to download Tidewave CLI"
+        rm -f "$temp_file"
+        return 1
+    fi
+}
+
 install_plugins() {
     export PATH="$HOME/.local/share/mise/shims:$HOME/.local/bin:$PATH"
     mkdir -p ~/.config/claude
@@ -1078,6 +1162,12 @@ install_plugins() {
     echo "Adding MCP servers..."
     claude mcp add playwright --scope user -- npx @anthropic-ai/mcp-server-playwright 2>/dev/null || true
     claude mcp add superpowers-chrome --scope user -- npx github:obra/superpowers-chrome --headless 2>/dev/null || true
+
+    # Note: Tidewave MCP is configured per-project when running a Phoenix app
+    # Use: claude mcp add --transport http tidewave http://localhost:4000/tidewave/mcp
+    echo ""
+    echo "Tidewave MCP: Add per-project with your Phoenix app running:"
+    echo "  claude mcp add --transport http tidewave http://localhost:PORT/tidewave/mcp"
 }
 
 install_base_deps() {
@@ -1115,8 +1205,29 @@ install_base_deps() {
         if command_exists cargo; then
             cargo install watchexec-cli
         else
-            # Fallback: install cargo first, then watchexec
+            # Security: Download rustup script first, validate, then execute (Issue #3)
-            curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh -s -- -y
+            # Same pattern as mise/ollama - never pipe directly to shell
+            local rustup_script
+            rustup_script=$(mktemp)
+            curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs -o "$rustup_script"
+
+            # Verify it's a shell script
+            if ! head -1 "$rustup_script" | grep -qE '^#!\s*/(bin|usr/bin)/(env\s+)?(ba)?sh'; then
+                log_error "Downloaded rustup script doesn't have a valid shell shebang"
+                rm -f "$rustup_script"
+                return 1
+            fi
+            # Check file size is reasonable (not empty, not huge)
+            local script_size
+            script_size=$(wc -c < "$rustup_script")
+            if [ "$script_size" -lt 100 ] || [ "$script_size" -gt 2000000 ]; then
+                log_error "Downloaded rustup script has suspicious size: $script_size bytes"
+                rm -f "$rustup_script"
+                return 1
+            fi
+
+            sh "$rustup_script" -y
+            rm -f "$rustup_script"
             source "$HOME/.cargo/env"
             cargo install watchexec-cli
         fi
@@ -1201,12 +1312,15 @@ register_step "base"       "System packages (git, build tools, curl)"
 register_step "postgresql" "PostgreSQL database server"
 register_step "mise"       "mise version manager"
 register_step "node"       "Node.js LTS runtime"
+register_step "python"     "Python (mise-managed)"
 register_step "erlang"     "Erlang/OTP VM"
 register_step "elixir"     "Elixir language"
 register_step "chromium"   "Chromium browser"
 register_step "vnc"        "VNC server + XFCE desktop"
 register_step "ollama"     "Ollama local LLM runner"
 register_step "claude"     "Claude Code CLI"
+register_step "opencode"   "OpenCode CLI"
+register_step "tidewave"   "Tidewave CLI (Elixir MCP)"
 register_step "playwright" "Playwright browser automation"
 register_step "plugins"    "Claude Code plugins"
 
@@ -1214,21 +1328,29 @@ register_step "plugins"    "Claude Code plugins"
 should_skip "postgresql" && skip_step "postgresql"
 should_skip "mise"       && skip_step "mise"
 should_skip "node"       && skip_step "node"
+should_skip "python"     && skip_step "python"
 should_skip "erlang"     && skip_step "erlang"
 should_skip "chromium"   && skip_step "chromium"
 should_skip "vnc"        && skip_step "vnc"
 should_skip "ollama"     && skip_step "ollama"
 should_skip "claude"     && skip_step "claude"
+should_skip "opencode"   && skip_step "opencode"
 should_skip "playwright" && skip_step "playwright"
 should_skip "plugins"    && skip_step "plugins"
 
 # Elixir skipped if Erlang skipped
 [ "${STEP_STATUS[erlang]}" = "skipped" ] && skip_step "elixir"
 
-# Claude/Playwright skipped if Node skipped
+# Tidewave skipped if Erlang skipped (it's for Phoenix/Elixir projects)
+[ "${STEP_STATUS[erlang]}" = "skipped" ] && skip_step "tidewave"
+
+# Python skipped if mise skipped
+[ "${STEP_STATUS[mise]}" = "skipped" ] && skip_step "python"
+
+# Claude/OpenCode/Playwright skipped if Node skipped
 if [ "${STEP_STATUS[node]}" = "skipped" ]; then
-    [ "${STEP_STATUS[mise]}" = "skipped" ] || true  # mise can still install
     skip_step "claude"
+    skip_step "opencode"
     skip_step "playwright"
     skip_step "plugins"
 fi
@@ -1267,19 +1389,23 @@ run_step_sync "base" install_base_deps || true
 wait_for_running_steps
 
 # ============================================================================
-# Phase 3: Node + Erlang (parallel, both need mise)
+# Phase 3: Node + Python + Erlang (parallel, all need mise)
 # ============================================================================
 if [ "${STEP_STATUS[mise]}" = "done" ]; then
     [ "${STEP_STATUS[node]}" != "skipped" ]   && start_step "node" install_node
+    [ "${STEP_STATUS[python]}" != "skipped" ] && start_step "python" install_python
     [ "${STEP_STATUS[erlang]}" != "skipped" ] && start_step "erlang" install_erlang
     wait_for_running_steps
 fi
 
 # ============================================================================
-# Phase 4: Elixir + Claude + Playwright (Elixir needs Erlang, others need Node)
+# Phase 4: Elixir + Claude + OpenCode + Playwright + Tidewave
+# (Elixir/Tidewave need Erlang, others need Node)
 # ============================================================================
 [ "${STEP_STATUS[erlang]}" = "done" ] && [ "${STEP_STATUS[elixir]}" != "skipped" ]    && start_step "elixir" install_elixir
+[ "${STEP_STATUS[erlang]}" = "done" ] && [ "${STEP_STATUS[tidewave]}" != "skipped" ]  && start_step "tidewave" install_tidewave
 [ "${STEP_STATUS[node]}" = "done" ] && [ "${STEP_STATUS[claude]}" != "skipped" ]      && start_step "claude" install_claude
+[ "${STEP_STATUS[node]}" = "done" ] && [ "${STEP_STATUS[opencode]}" != "skipped" ]    && start_step "opencode" install_opencode
 [ "${STEP_STATUS[node]}" = "done" ] && [ "${STEP_STATUS[playwright]}" != "skipped" ]  && start_step "playwright" install_playwright
 
 wait_for_running_steps
@@ -1354,6 +1480,12 @@ fi
 if [ "${STEP_STATUS[claude]}" = "done" ]; then
     echo -e "${CYAN}Claude:${NC}      claude"
 fi
+if [ "${STEP_STATUS[opencode]}" = "done" ]; then
+    echo -e "${CYAN}OpenCode:${NC}    opencode"
+fi
+if [ "${STEP_STATUS[tidewave]}" = "done" ]; then
+    echo -e "${CYAN}Tidewave:${NC}    tidewave (run in Phoenix project dir)"
+fi
 if [ "${STEP_STATUS[ollama]}" = "done" ]; then
     echo -e "${CYAN}Ollama:${NC}      ollama run llama3.2"
 fi