md-review

name	md-review
description	Use when design document needs user review - auto-opens browser to render MD content
trigger	context-based (LLM decides when to invoke)
author	skills-team

MD Review

Automatically opens a browser to render Markdown design documents for review.

When to Use

• User needs to review a design document
• LLM detects discussion about a design spec
• File path extracted from conversation context

How It Works

1. LLM calls skill with file path (via parameter, context, or config pattern)
2. Skill generates temporary HTML with GitHub-style rendering
3. Python HTTP server starts on random available port
4. Platform-appropriate command opens browser
5. User reviews in browser; files cleaned on session end

File Specification (Priority)

1. file: parameter - explicit path passed to skill
2. Conversation context - LLM extracts from current discussion
3. Config patterns - glob matching in config.json

Supported Platforms

Platform	Browser Command
Linux/WSL	xdg-open
macOS	open
Windows	start

Usage

# Direct invocation (LLM handles this)
./scripts/server.sh start /path/to/design.md

# The skill is invoked by LLM when context indicates review needed

Cleanup

Server and temp files are cleaned when Claude Code session ends. PID tracked in ~/.claude/md-review.pid