Dashboard: MCP tool coverage gaps — 26 tables with no tools

[erikdarlingdata]

Problem

26 of 52 Dashboard database tables have no MCP tool coverage. Major categories missing:

High Priority

Table	What it contains
collect.query_snapshots_*	Active queries (sp_WhoIsActive daily tables)
collect.cpu_scheduler_stats	Scheduler pressure, worker threads, runnable tasks
collect.latch_stats	Latch contention by class
collect.spinlock_stats	Spinlock collisions and backoffs
collect.waiting_tasks	Currently waiting tasks with blocking chains
collect.session_stats	Connection counts, dormant sessions
collect.plan_cache_stats	Plan cache composition, single-use bloat

Medium Priority

Table	What it contains
collect.database_size_stats	Database file sizes
collect.server_properties	Server inventory (edition, CPU, memory)
collect.memory_pressure_events	Ring buffer memory pressure notifications
collect.default_trace_events	Autogrow, config changes, memory errors
collect.trace_analysis	Long-running query trace data
config.server_configuration_history	sp_configure change history
config.database_configuration_history	Database config change history
config.trace_flags_history	Trace flag change history
config.critical_issues	Detected performance issues

Lower Priority (11 HealthParser tables)

All collect.HealthParser_* tables (CPUTasks, IOIssues, MemoryBroker, MemoryConditions, MemoryNodeOOM, SchedulerIssues, SevereErrors, SignificantWaits, SystemHealth, WaitsByCount, WaitsByDuration) — parsed system_health extended event data.

Note

Raw XML staging tables (deadlock_xml, blocked_process_xml) are low priority since their parsed equivalents (deadlocks, blocking_BlockedProcessReport) already have MCP coverage.

Acceptance Criteria

• High priority tables each get a dedicated MCP tool
• Tools support server_name and hours_back parameters
• Dashboard has reporting views for most of these already — tools should use them