panopticon-cli 0.4.0 → 0.4.4

package/README.md

@@ -1,33 +1,72 @@
+<div align="center">
+
 # Panopticon CLI
 
-Multi-agent orchestration for AI coding assistants.
+**Multi-agent orchestration for AI coding assistants**
+
+[![npm version](https://img.shields.io/npm/v/panopticon-cli.svg)](https://www.npmjs.com/package/panopticon-cli)
+[![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](https://opensource.org/licenses/MIT)
+[![Node.js Version](https://img.shields.io/badge/node-%3E%3D18-brightgreen.svg)](https://nodejs.org/)
+[![PRs Welcome](https://img.shields.io/badge/PRs-welcome-brightgreen.svg)](https://github.com/eltmon/panopticon/pulls)
 
 > *"The Panopticon had six sides, one for each of the Founders of Gallifrey..."*
 
-## Overview
+<img src="docs/dashboard-overview.png" alt="Panopticon Dashboard" width="800" />
+
+</div>
+
+---
 
-Panopticon is a unified orchestration layer for AI coding assistants. It works with:
+## What is Panopticon?
+
+| Without Panopticon | With Panopticon |
+|:------------------|:----------------|
+| Manually juggle multiple AI agents | **Automatic orchestration** - spawn, monitor, and coordinate agents from a dashboard |
+| Agents start fresh every session | **Persistent context** - skills, state files, and beads track work across sessions |
+| Simple tasks eat Opus credits | **Smart model routing** - Haiku for simple, Sonnet for medium, Opus for complex |
+| Stuck agents waste your time | **Automatic recovery** - detect stuck agents and hand off to specialists |
+| AI tools have separate configs | **Universal skills** - one SKILL.md works across Claude, Codex, Cursor, Gemini |
+
+## Screenshots
+
+<div align="center">
+<table>
+<tr>
+<td><img src="docs/planning-session-dialog.png" alt="Planning Dialog" width="300" /></td>
+<td><img src="docs/planning-session-discovery.png" alt="Discovery Phase" width="300" /></td>
+<td><img src="docs/planning-session-active.png" alt="Active Session" width="300" /></td>
+</tr>
+<tr>
+<td align="center"><em>Start planning</em></td>
+<td align="center"><em>Discovery phase</em></td>
+<td align="center"><em>Active session</em></td>
+</tr>
+</table>
+</div>
+
+## Key Features
+
+| Feature | Description |
+|:--------|:------------|
+| **Multi-Agent Orchestration** | Spawn and manage AI agents in tmux sessions via dashboard or CLI |
+| **Cloister Lifecycle Manager** | Automatic model routing, stuck detection, and specialist handoffs |
+| **Universal Skills** | One SKILL.md format works across all supported AI tools |
+| **Workspaces** | Git worktree-based feature branches with Docker isolation |
+| **Convoys** | Run parallel agents on related issues with auto-synthesis |
+| **Specialists** | Dedicated review, test, and merge agents for quality control |
+| **Heartbeat Monitoring** | Real-time agent activity tracking via Claude Code hooks |
+| **Legacy Codebase Support** | AI self-monitoring skills that learn from your codebase |
+
+## Supported Tools
 
 | Tool | Support |
-|------|---------|
+|:-----|:--------|
 | **Claude Code** | Full support |
 | **Codex** | Skills sync |
 | **Cursor** | Skills sync |
 | **Gemini CLI** | Skills sync |
 | **Google Antigravity** | Skills sync |
 
-### Features
-
-- **Multi-agent orchestration** - Spawn and manage multiple AI agents in tmux sessions
-- **Cloister AI Lifecycle Manager** - Automatic model routing, stuck detection, and specialist handoffs
-- **Universal skills** - One SKILL.md format works across all supported tools
-- **Heartbeat Hooks** - Real-time agent activity monitoring via Claude Code hooks
-- **Multi-project support** - YAML-based project registry with label-based issue routing
-- **Health Monitoring** - Deacon-style stuck detection with auto-recovery
-- **Context Engineering** - Structured state management (STATE.md, WORKSPACE.md)
-- **Agent CVs** - Work history tracking for capability-based routing
-- **Google Stitch Integration** - AI-powered UI design with automatic design-to-code workflows
-
 ---
 
 ## Legacy Codebase Support
@@ -159,54 +198,12 @@ The AI adapts to your org structure, not the other way around.
 ## Quick Start
 
 ```bash
-# Install Panopticon
-npm install -g panopticon-cli
-
-# Install prerequisites and setup (includes optional HTTPS/Traefik)
-pan install
-
-# Sync skills to all AI tools
-pan sync
-
-# Check system health
-pan doctor
-```
-
-### HTTPS Setup (Optional)
-
-Panopticon supports local HTTPS via Traefik reverse proxy:
-
-```bash
-# Full install (includes Traefik + mkcert for HTTPS)
-pan install
-
-# Add to /etc/hosts (macOS/Linux)
-echo "127.0.0.1 pan.localhost" | sudo tee -a /etc/hosts
-
-# Start with HTTPS
-pan up
-# → Dashboard: https://pan.localhost
-# → Traefik UI: https://traefik.pan.localhost:8080
-```
-
-**Minimal install** (skip Traefik, use ports):
-```bash
-pan install --minimal
-pan up
-# → Dashboard: http://localhost:3010
+npm install -g panopticon-cli && pan install && pan sync && pan up
 ```
 
-See [docs/DNS_SETUP.md](docs/DNS_SETUP.md) for detailed DNS configuration (especially for WSL2).
-
-## Supported Platforms
+**That's it!** Dashboard runs at https://pan.localhost (or http://localhost:3010 if you skip HTTPS setup).
 
-| Platform | Support |
-|----------|---------|
-| **Linux** | Full support |
-| **macOS** | Full support |
-| **Windows** | WSL2 required |
-
-> **Windows users:** Panopticon requires WSL2 (Windows Subsystem for Linux 2). Native Windows is not supported. [Install WSL2](https://docs.microsoft.com/en-us/windows/wsl/install)
+📖 **[Full installation guide, requirements, and platform support →](https://panopticon-cli.com/quickstart)**
 
 ## Requirements
 
@@ -270,6 +267,111 @@ Panopticon supports multiple issue trackers:
 
 Secondary trackers sync issues to the dashboard alongside Linear issues, allowing unified project management.
 
+### Multi-Model Support
+
+Panopticon integrates with [claude-code-router](https://github.com/musistudio/claude-code-router) to enable using multiple AI model providers alongside Anthropic models. Configure which models to use for different agent types and task complexities.
+
+📖 **[Complete work types guide and configuration examples →](docs/WORK-TYPES.md)**
+📋 **[Configuration file reference and presets →](docs/CONFIGURATION.md)**
+🧠 **[Model recommendations for optimal cost/performance →](docs/MODEL_RECOMMENDATIONS.md)**
+
+#### Supported Providers and Models
+
+**Anthropic** (via Claude Code / Claude API)
+- `claude-opus-4-5` - Most capable, best for planning and complex tasks
+- `claude-sonnet-4-5` - Balanced performance and cost
+- `claude-haiku-4-5` - Fast and cost-effective for simple tasks
+
+**OpenAI**
+- `gpt-5.2-codex` - Agentic coding and research
+- `o3-deep-research` - Deep research capabilities
+- `gpt-4o` - General-purpose model
+- `gpt-4o-mini` - Faster, more cost-effective variant
+
+**Google (Gemini)**
+- `gemini-3-pro-preview` - Supports `thinking_level: high|low`
+- `gemini-3-flash-preview` - Supports `thinking_level: minimal|low|medium|high`
+
+**Z.AI**
+- `glm-4.7` - General-purpose model
+- `glm-4.7-flash` - Faster variant
+
+#### Configuration via Dashboard
+
+1. Open the Panopticon dashboard and navigate to **Settings**
+2. Configure **API keys** for external providers:
+   - OpenAI API Key
+   - Google AI API Key
+   - Z.AI API Key
+3. Configure **models per agent type**:
+   - Review Agent - Model for code review
+   - Test Agent - Model for running tests
+   - Merge Agent - Model for handling merges
+   - Planning Agent - Model for autonomous planning
+4. Configure **models by task complexity** (for PAN-75):
+   - Trivial tasks → Cost-effective models (e.g., `claude-haiku-4-5`)
+   - Expert tasks → Most capable models (e.g., `claude-opus-4-5`)
+
+**Configuration Files:**
+
+Panopticon uses YAML-based configuration (new in v0.5+):
+
+| File | Purpose |
+|------|---------|
+| `~/.panopticon/config.yaml` | Global model settings, provider enable/disable, API key references |
+| `~/.panopticon.env` | API keys and sensitive credentials (auto-loaded) |
+| `.panopticon.yaml` | Per-project config (optional, overrides global) |
+
+**Security:** For added security, restrict file permissions:
+```bash
+chmod 600 ~/.panopticon/config.yaml ~/.panopticon.env
+```
+
+**Environment Variables:** API keys are loaded from `~/.panopticon.env` at startup:
+```bash
+# ~/.panopticon.env
+KIMI_API_KEY="sk-kimi-..."
+OPENAI_API_KEY="sk-..."
+GOOGLE_AI_KEY="AIza..."
+ZAI_API_KEY="..."
+```
+
+In `config.yaml`, reference them with `$` syntax:
+```yaml
+models:
+  providers:
+    kimi:
+      enabled: true
+      api_key: $KIMI_API_KEY
+```
+
+#### Router Configuration
+
+**Panopticon owns the router configuration.** When you save settings in the dashboard, Panopticon automatically generates `~/.claude-code-router/config.json` with the appropriate provider and routing configuration. Manual edits to this file will be overwritten on the next settings change.
+
+#### Example Use Cases
+
+**Cost optimization:**
+```
+Specialist agents → claude-sonnet-4-5 (balanced)
+Planning agent → claude-opus-4-5 (most capable)
+Simple tasks → claude-haiku-4-5 (cost-effective)
+```
+
+**Multi-provider setup:**
+```
+Code review → gpt-4o (OpenAI's code understanding)
+Testing → claude-sonnet-4-5 (Anthropic's reliability)
+Planning → claude-opus-4-5 (Anthropic's reasoning)
+```
+
+**Research-heavy workflows:**
+```
+Research tasks → o3-deep-research (OpenAI)
+Implementation → claude-sonnet-4-5 (Anthropic)
+Code review → gemini-3-pro-preview (Google)
+```
+
 ### Register Projects
 
 Register your local project directories so Panopticon knows where to create workspaces:
@@ -347,6 +449,35 @@ When Stitch is configured, the planning agent can:
 2. **Document in STATE.md** - Record Stitch project/screen IDs for workers
 3. **Worker agents pick up designs** - Use the Stitch skills to convert to React
 
+### Bidirectional Sync
+
+Designs are stored in your Google account and sync bidirectionally:
+
+```
+Planning Agent creates design via MCP
+    ↓
+Stored in Google Cloud (your account)
+    ↓
+Visible on stitch.withgoogle.com
+    ↓
+Worker agent retrieves via MCP → converts to React
+```
+
+- **MCP → Web UI:** Designs created by agents appear in Stitch web UI
+- **Web UI → MCP:** Designs created in web UI are accessible to agents
+- **Edit anywhere:** Refine AI-generated designs in the web UI before workers convert them
+
+### Available MCP Tools
+
+| Tool | Purpose |
+|------|---------|
+| `create_project` | Create a new Stitch project |
+| `list_projects` | List all your projects (owned/shared) |
+| `get_project` | Get project details |
+| `list_screens` | List screens in a project |
+| `get_screen` | Get screen details + HTML/CSS download URLs |
+| `generate_screen_from_text` | Generate a screen from a text prompt |
+
 ### Stitch Skills
 
 | Skill | Purpose |
@@ -1054,7 +1185,50 @@ Panopticon creates workspaces as git worktrees. Docker, HTTPS, and seeding are o
 
 ## Polyrepo Workspace Configuration
 
-For projects with multiple git repositories (like separate frontend/backend repos), configure workspace settings directly in `projects.yaml`:
+For projects with multiple git repositories (like separate frontend/backend repos), configure workspace settings directly in `projects.yaml`.
+
+### Recommended: Separate Infra Repository
+
+For polyrepo projects, we strongly recommend maintaining a **dedicated infrastructure repository** alongside your code repos:
+
+```
+myproject/
+├── infra/                    # Infrastructure repo (git)
+│   ├── .devcontainer-template/   # Templates for workspace creation
+│   │   ├── dev.template          # Dev script template
+│   │   ├── compose.infra.yml.template
+│   │   └── docker-compose.devcontainer.yml.template
+│   ├── new-feature              # Script to create workspaces
+│   ├── remove-feature           # Script to clean up workspaces
+│   ├── certs/                   # SSL certificates
+│   └── traefik/                 # Traefik config
+├── frontend/                 # Frontend repo (git)
+├── backend/                  # Backend repo (git)
+└── workspaces/               # Feature workspaces (git worktrees)
+    └── feature-xxx/
+        ├── fe/               # Worktree of frontend
+        ├── api/              # Worktree of backend
+        └── .devcontainer/    # Generated from templates
+```
+
+**Benefits of a separate infra repo:**
+
+1. **Version-controlled templates** - Dev script changes are tracked and shared
+2. **Centralized configuration** - DNS, ports, Traefik rules in one place
+3. **Workspace consistency** - All workspaces use the same templates
+4. **Team collaboration** - Infra changes go through normal PR review
+5. **Easy updates** - Fix a template once, regenerate affected workspaces
+
+**Key files in the infra repo:**
+
+| File | Purpose |
+|------|---------|
+| `dev.template` | Template for `./dev` script (container management) |
+| `new-feature` | Creates workspace with git worktrees + devcontainer |
+| `remove-feature` | Cleans up workspace and worktrees |
+| `.devcontainer-template/` | Docker Compose and devcontainer templates |
+
+
 
 ```yaml
 projects:
@@ -1468,6 +1642,165 @@ pan work recover min-123
 pan work recover --all
 ```
 
+### Convoys: Multi-Agent Orchestration
+
+Convoys enable Panopticon to run multiple AI agents in parallel for complex tasks like code review. Instead of a single agent doing everything, specialized agents focus on specific concerns and a synthesis agent combines their findings.
+
+**Why Convoys?**
+
+When reviewing code, a single AI agent must context-switch between:
+- Checking for logic errors
+- Looking for security vulnerabilities
+- Analyzing performance issues
+
+This leads to:
+- Shallow reviews (can't go deep on everything)
+- Missed issues (focus on one area, miss others)
+- Long sequential execution (can't parallelize)
+
+Convoys solve this by **specialization + parallelization**:
+- 3 focused agents review in parallel (10x faster)
+- Each agent goes deep in their domain
+- Synthesis agent combines findings with prioritization
+
+#### Built-in Convoy Templates
+
+| Template | Agents | Use Case |
+|----------|--------|----------|
+| `code-review` | correctness, security, performance, synthesis | Comprehensive code review |
+| `planning` | planner | Codebase exploration and planning |
+| `triage` | (dynamic) | Parallel issue triage |
+| `health-monitor` | monitor | Check health of running agents |
+
+#### Quick Start
+
+```bash
+# Run a parallel code review
+pan convoy start code-review --files "src/**/*.ts"
+
+# Check status
+pan convoy status
+
+# List all convoys (running and completed)
+pan convoy list
+
+# Stop a convoy
+pan convoy stop <convoy-id>
+```
+
+#### How Code Review Convoy Works
+
+```
+pan convoy start code-review --files "src/**/*.ts"
+    │
+    ├─→ Phase 1 (Parallel): 3 specialized reviewers run simultaneously
+    │     ├─→ Correctness (Haiku) → correctness.md
+    │     ├─→ Security (Sonnet)    → security.md
+    │     └─→ Performance (Haiku)  → performance.md
+    │
+    └─→ Phase 2 (Sequential): Synthesis agent runs after Phase 1 completes
+          └─→ Synthesis reads all 3 reviews → synthesis.md
+```
+
+**Phase 1 - Specialized Reviews (Parallel)**
+- **Correctness** (Haiku) - Logic errors, edge cases, type safety
+- **Security** (Sonnet) - OWASP Top 10, injection, XSS, auth issues
+- **Performance** (Haiku) - N+1 queries, blocking ops, memory leaks
+
+**Phase 2 - Synthesis (Sequential)**
+- Reads all three review files
+- Removes duplicates (same issue found by multiple reviewers)
+- Prioritizes findings (blocker → critical → high → medium → low)
+- Generates unified report with action items
+
+#### What is Synthesis?
+
+Synthesis is the process of combining findings from multiple parallel agents into a single, prioritized, actionable report.
+
+**Without synthesis**, after 3 parallel reviews you get:
+- 3 separate markdown files to read
+- Duplicate findings (same issue reported differently)
+- No prioritization (which to fix first?)
+- Mental overhead to merge them yourself
+
+**With synthesis**, you get:
+- Single unified report
+- Deduplicated findings
+- AI-prioritized by severity × impact
+- Clear action items
+
+**Example synthesis output:**
+
+```markdown
+# Code Review - Complete Analysis
+
+## Executive Summary
+- 2 blockers (MUST FIX)
+- 3 critical issues
+- 5 high-priority items
+
+## Top Priority
+1. SQL injection in auth.ts:42 (Security, Critical)
+2. execSync blocking event loop in sync.ts:89 (Performance, Blocker)
+3. Null pointer in user fetch in api.ts:156 (Correctness, High)
+
+## Blocker Issues
+[Detailed findings with code examples and fixes]
+
+## Critical Issues
+[...]
+
+## Review Statistics
+- Files reviewed: 12
+- Issues found: 10
+- Duplicates removed: 3
+```
+
+#### Convoy Commands
+
+```bash
+# Start a convoy
+pan convoy start <template> [options]
+  --files <pattern>       # File pattern (e.g., "src/**/*.ts")
+  --pr-url <url>          # Pull request URL
+  --issue-id <id>         # Issue ID
+  --project-path <path>   # Project path (defaults to cwd)
+
+# Check convoy status
+pan convoy status [convoy-id]  # Show status (defaults to most recent)
+
+# List convoys
+pan convoy list                # All convoys
+pan convoy list --status running  # Filter by status
+
+# Stop a convoy
+pan convoy stop <convoy-id>    # Kill all agents, mark failed
+```
+
+#### Custom Convoy Templates
+
+Create custom templates in `~/.panopticon/convoy-templates/`:
+
+```json
+{
+  "name": "lightweight-review",
+  "description": "Quick security-only review",
+  "agents": [
+    {
+      "role": "security",
+      "subagent": "code-review-security",
+      "parallel": false
+    }
+  ],
+  "config": {
+    "outputDir": ".claude/reviews",
+    "timeout": 600000
+  }
+}
+```
+
+Then use with: `pan convoy start lightweight-review --files "src/**/*.ts"`
+
 ### Health Monitoring
 
 ```bash
@@ -1629,15 +1962,24 @@ These safeguards are enforced through:
 | ![Start](docs/planning-session-dialog.png) | ![Exploring](docs/planning-session-active.png) | ![Discovery](docs/planning-session-discovery.png) |
 | Click to create workspace and start AI planning | Claude explores the codebase, reads docs, understands patterns | Interactive questions to clarify requirements and approach |
 
-Planning artifacts are stored **inside the workspace**, making them part of the feature branch:
+Planning artifacts are stored **inside the workspace**:
+
+> ⚠️ **Note: `.planning/` is currently ephemeral (not tracked in git)**
+>
+> The `.planning/` directory is listed in `.gitignore` to prevent cross-contamination between issues. When branches merge, planning state from issue A could contaminate issue B's workspace, causing agents to work on the wrong tasks.
+>
+> **Current behavior:** Planning state is local to your machine only. If you switch machines, you'll need to re-run the planning session.
+>
+> See [#111](https://github.com/eltmon/panopticon-cli/issues/111) for the future enhancement to support cross-machine planning sync safely.
 
 ```
 workspaces/feature-min-123/
-├── .planning/
-│   ├── output.jsonl          # Full conversation history (tool uses + results)
-│   ├── PLANNING_PROMPT.md    # Initial planning prompt
+├── .planning/                 # ⚠️ NOT tracked in git (ephemeral)
+│   ├── STATE.md               # Current planning state
+│   ├── output.jsonl           # Full conversation history
+│   ├── PLANNING_PROMPT.md     # Initial planning prompt
 │   ├── CONTINUATION_PROMPT.md # Context for continued sessions
-│   └── output-*.jsonl        # Backup of previous rounds
+│   └── output-*.jsonl         # Backup of previous rounds
 └── ... (code)
 ```
 
@@ -1648,42 +1990,16 @@ When the planning session completes, the AI generates:
 - **Beads tasks** - Trackable sub-tasks with dependencies for the implementation
 - **Feature PRD** - Copied to `docs/prds/active/{issue}-plan.md` for documentation
 
-**This enables:**
-
-1. **Collaborative async planning** - Push your branch, someone else pulls and continues the planning session with full context
-2. **Context recovery** - If Claude's context compacts, the full conversation is preserved in the branch
-3. **Audit trail** - See how planning decisions were made, what files were explored, what questions were asked
-4. **Branch portability** - The planning state travels with the feature branch
+**Future capabilities (currently disabled - see note above):**
 
-**Dashboard workflow (recommended):**
+1. ~~**Collaborative async planning** - Push your branch, someone else pulls and continues the planning session with full context~~
+2. ~~**Branch portability** - The planning state travels with the feature branch~~
 
-The planning dialog has **Pull** and **Push** buttons that handle git operations automatically:
+**What still works:**
 
-| Button | What it does |
-|--------|--------------|
-| **Pull** | Fetches from origin, creates workspace from remote branch if needed, pulls latest changes |
-| **Push** | Commits `.planning/` artifacts and pushes to origin |
-
-1. Person A starts planning in dashboard, clicks **Push** when interrupted
-2. Person B opens same issue in dashboard, clicks **Pull** → gets Person A's full context
-3. Person B continues the planning session and clicks **Push** when done
-
-**CLI workflow:**
-```bash
-# Person A starts planning
-pan work plan MIN-123
-# ... answers discovery questions, gets interrupted ...
-
-# Push the branch (includes planning context)
-cd workspaces/feature-min-123
-git add .planning && git commit -m "WIP: planning session"
-git push origin feature/min-123
-
-# Person B continues
-git pull origin feature/min-123
-pan work plan MIN-123 --continue
-# Claude has full context from Person A's session
-```
+1. **Context recovery** - If Claude's context compacts, STATE.md and beads preserve the planning decisions
+2. **Audit trail** - See decisions made in STATE.md
+3. **Beads tasks** - Sub-tasks created during planning ARE persisted (beads uses label-based isolation)
 
 ```bash
 # Create a workspace (git worktree) without starting an agent
@@ -2472,15 +2788,26 @@ localhostForwarding=true
 # Disable GPU/GUI passthrough - reduces dxg driver errors
 guiApplications=false
 
+# ============================================================
+# WINDOWS 11 ONLY - Comment these out on Windows 10!
+# These settings require Windows 11 22H2 or later.
+# On Windows 10 they are silently ignored or may cause issues.
+# ============================================================
+
 # Route DNS through Windows - prevents getaddrinfo() failures
-dnsTunneling=true
+# dnsTunneling=true          # Windows 11 only
 
 # Inherit Windows proxy settings
-autoProxy=true
+# autoProxy=true             # Windows 11 only
 
-# Windows 11 22H2+ ONLY: Use mirrored networking
-# This bypasses the Hyper-V NAT layer that can cause "Object Name not found" errors
-# networkingMode=mirrored
+# Aggressively reclaim cached memory
+# autoMemoryReclaim=dropCache  # Windows 11 only
+
+# Use sparse VHD to allow better memory compaction
+# sparseVhd=true             # Windows 11 only
+
+# Mirrored networking (may conflict with Docker Desktop)
+# networkingMode=mirrored    # Windows 11 only
 ```
 
 **After changing `.wslconfig`:**
@@ -2505,15 +2832,21 @@ ip addr show eth0  # NAT mode shows 172.x.x.x IP
 
 | Feature | Windows 10 | Windows 11 22H2+ |
 |---------|-----------|------------------|
-| `networkingMode=mirrored` | ❌ Not supported | ✅ Supported |
-| `dnsTunneling=true` | ✅ Works | ✅ Works |
+| `memory`, `processors`, `swap` | ✅ Works | ✅ Works |
+| `localhostForwarding` | ✅ Works | ✅ Works |
 | `guiApplications=false` | ✅ Works | ✅ Works |
+| `networkingMode=mirrored` | ❌ Not supported | ✅ Supported |
+| `dnsTunneling=true` | ❌ Not supported | ✅ Supported |
+| `autoProxy=true` | ❌ Not supported | ✅ Supported |
+| `autoMemoryReclaim` | ❌ Not supported | ✅ Supported |
+| `sparseVhd=true` | ❌ Not supported | ✅ Supported |
 
-**Windows 10 users:** The `networkingMode=mirrored` setting will be silently ignored. You'll stay on NAT mode, which can be less stable. The other settings (`dnsTunneling`, `guiApplications=false`) still help with stability.
+**Windows 10 users:** Most advanced WSL2 features require Windows 11. On Windows 10, only the basic resource limits and `localhostForwarding`/`guiApplications` settings are supported. Other settings will be silently ignored or may cause instability.
 
 If you experience frequent WSL2 crashes on Windows 10, consider:
-- Upgrading to Windows 11 for mirrored networking support
+- Using only the basic settings shown above (memory, processors, swap, localhostForwarding, guiApplications)
 - Reducing `memory` allocation if system is under pressure
+- Upgrading to Windows 11 for full WSL2 feature support
 - Checking Windows Event Viewer for specific crash causes
 
 #### Additional Windows 10 Workarounds

package/dist/{agents-RCAPFJG7.js → agents-B5NRTVHK.js}

@@ -16,9 +16,8 @@ import {
   saveSessionId,
   spawnAgent,
   stopAgent
-} from "./chunk-4BIZ4OVN.js";
-import "./chunk-U4LCHEVU.js";
-import "./chunk-DGUM43GV.js";
+} from "./chunk-H45CLB7E.js";
+import "./chunk-7HHDVXBM.js";
 export {
   appendActivity,
   autoRecoverAgents,
@@ -38,4 +37,4 @@ export {
   spawnAgent,
   stopAgent
 };
-//# sourceMappingURL=agents-RCAPFJG7.js.map
+//# sourceMappingURL=agents-B5NRTVHK.js.map