Observability and Health Checks

Uptime monitoring cannot see the way remote MCP servers actually break

A dev.to writeup makes the structural case: tool and schema drift exists only as a diff between two snapshots, so no single health probe can find it, however thorough.