pg-health-check

PostgreSQL Health Check

This skill guides a comprehensive health assessment of a PostgreSQL database using pgtuner-mcp tools, producing a scored report with prioritized action items.

When to Use This Skill

Use this skill when the user:

• Asks for a database health check or status report
• Reports general performance degradation without specific symptoms
• Wants a routine database checkup or audit
• Asks "is my database healthy?" or "what should I fix first?"
• Is setting up monitoring or wants to establish a baseline
• Mentions connection issues, high CPU, or unexplained slowness

MCP Resources Available

Before running full tool scans, the agent can use targeted resources:

• pgtuner://health/{check_type} -- Targeted health checks: connections, cache, locks, replication, bloat, all (lightweight, avoids running the full health tool)
• pgtuner://settings/{category} -- Settings by category: memory, checkpoint, wal, autovacuum, connections
• pgtuner://docs/workflows -- Recommended workflow patterns

Smart resource usage: After the full health check in Step 1, use targeted resources for follow-up instead of re-running the full tool. For example, if the locks dimension scores low, read pgtuner://health/locks for quick re-checks after remediation.

Related MCP Prompt

This skill corresponds to the health_check MCP Prompt. If the user triggers that prompt, follow this skill's workflow.

Prerequisites

• The pgtuner-mcp server must be connected with a valid DATABASE_URI
• Required: pg_stat_statements extension
• Recommended: monitoring user with access to pg_stat_activity, pg_stat_bgwriter, pg_locks

Agent Decision Tree

User asks for health check
  |
  +--> Quick check or full check?
  |    QUICK --> Read pgtuner://health/all resource (fast, no tool call)
  |    FULL  --> Start with Step 1 (check_database_health)
  |
  +--> After Step 1, which dimensions scored low?
       |
       +--> Connections < 90 --> Prioritize Step 2, then pg-connection-analysis skill
       +--> Cache < 90      --> Prioritize Step 5 (I/O), then pg-io-deep-dive skill
       +--> Locks < 90      --> Prioritize Step 2 + Step 3, then pg-lock-diagnosis skill
       +--> Wraparound < 90 --> URGENT: check pg-vacuum-tuning skill
       +--> Checkpoints < 90 --> Prioritize Step 5, then pg-config-tuning skill
       +--> All >= 90       --> Summary report, suggest pg-performance-baseline