Add asset-health skill

[linazhan]

Summary

• New asset-health skill that checks the health of a data table via Monte Carlo MCP tools
• Produces a structured report with: last activity, active alerts, monitoring coverage, importance, upstream dependency health, and Monte Carlo links
• Wired into all editor plugins (claude-code, cursor, opencode, codex, copilot) via symlinks

Skill structure

skills/asset-health/
├── SKILL.md                    # Triggers, output format, metric definitions
├── README.md                   # Setup and usage guide
└── references/
    ├── workflows.md            # Phased workflow with parallel execution
    └── parameters.md           # MCP tool parameters and field mappings

Key design decisions

• Read-only informational skill — no hooks, no gates
• 1-hop upstream check only; asks user before drilling deeper
• Caps upstream parent checks at 10
• Uses get_mc_webapp_url for environment-specific link construction
• All sections always shown (with empty-state text if no data)
• Existing MCP tools sufficient — no backend changes needed

Test plan

• Tested manually with Claude Code against live Monte Carlo instance
• Test with Cursor plugin installation
• Verify skill triggers correctly ("how is table X", "check on X", "status of X")
• Verify disambiguation table shows warehouse column for multiple results
• Verify alert table respects 5-alert limit with overflow message

[linear]

DOL-8624 Add 'data-health' skill