More screenshots

    Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIMuseMuseDotsDots+21 more

    PostgreSQL Operations Diagnostics

    1

    Give your AI agent structured PostgreSQL investigation workflows through your own connected MCP-PostgreSQL-Ops server.

    Free

    Security scanned

    About this skill

    Give your AI agent structured PostgreSQL investigation workflows through your own connected MCP-PostgreSQL-Ops server.

    This skill helps your agent select the relevant tools, establish the target database, gather evidence and explain what the results actually prove. It covers:

    • Server identity, database inventory, active sessions, locks and wait events.
    • Slow-query statistics when pg_stat_statements or pg_stat_monitor is available.
    • Database, table and index I/O, buffer usage and checkpoint statistics.
    • Storage usage, dead-tuple-based bloat estimates and autovacuum investigation.
    • Running maintenance, last maintenance timestamps and cumulative counters.
    • Schema structure, declared relationships, role information and focused settings.
    • WAL, replication, slots and standby conflict inspection.
    • Version-specific diagnostics, including PostgreSQL 17 wait-event catalogs and WAL summarizer inspection, and PostgreSQL 18 asynchronous/per-backend I/O tools.

    The self-contained skill includes an individual operating guide for all 38 registered tools: 37 database inspection tools and one bundled-guidance tool. Each entry gives its source-matched signature, purpose, evidence and interpretation limits. Six symptom-driven playbooks connect those tools for health reviews, blocking, slow queries and I/O, bloat and autovacuum, schema review, and WAL and replication investigations. Illustrative calls show valid argument usage without presenting invented database results.

    The instructions distinguish estimates from measurements, current snapshots from historical evidence, and recommendations from executed changes. They account for missing permissions, optional extensions, statistics resets, version-dependent fallbacks and partial tool results.

    All workflows use inspection tools. The skill does not run VACUUM, terminate sessions, alter settings, create or drop indexes, install extensions, restart PostgreSQL or perform failover. It helps you prepare informed maintenance decisions; execution requires a separate authorized workflow.

    Read-only investigation only. No support for executing maintenance, terminating sessions, or changing configuration. Requires pre-configured MCP server and database permissions.

    How to use

    1. 1

      Connect your agent to the Agensi MCP

      Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup

    2. 2

      Sign in to Agensi

      Your agent shows a sign-in link. Sign in and click Allow.

    3. 3

      Ask for what you need

      Your agent loads this skill and uses it on your task. It always asks before any payment.

    Reviews

    No reviews yet

    Be one of the first to try it. Every listed skill goes through the trust checks below.

    Security scanned

    Checked by our 8-point scan before listing

    Fresh listing

    Recently published to Agensi

    Free forever

    No account required to browse

    Trust & safety

    Security scan

    Security scan passed 19 days ago

    • Free with an Agensi account

    Requires

    PostgreSQL Ops MCP
    PostgreSQL 12–18
    Python 3.12+

    Permissions

    Network Access

    Uses the buyer's connected MCP-PostgreSQL-Ops server to inspect their PostgreSQL instance. The server operator supplies database endpoints, credentials, network access and appropriate monitoring permissions. These workflows query database state and statistics; they do not change database objects, settings, sessions or maintenance state. No agent access to local files, shell, browser or environment variables is required. Database traffic originates from the MCP server. Server-side connection configuration is not an agent permission grant. Query text, settings, role information and client addresses may be sensitive; collect only necessary evidence and sanitize shared output. The reviewed server can log database credentials at DEBUG level; avoid DEBUG logging and protect existing logs.

    Listed19 days ago

    What's inside

    Frequently Asked Questions

    No. Connect your own MCP-PostgreSQL-Ops server to your own PostgreSQL environment. This product contains agent instructions only.

    Core catalog diagnostics do not require them. The query-statistics tools need their respective extensions and configuration. Managed services may restrict extension availability. Extension checks in the reviewed wrappers use the default database, which can differ from the requested target database.

    No. It provides read-only diagnosis and recommendations. Database mutations and maintenance execution are outside this skill. The server's registered tool set does not expose arbitrary SQL or maintenance commands. Use a restricted database role: agent instructions do not enforce server-side privileges.

    No. It is a dead-tuple-based estimate, not physical page inspection. Ordinary VACUUM does not guarantee shrinking database files on disk.

    No. Most statistics are current snapshots or cumulative counters since reset. Autovacuum activity uses last-run timestamps and totals, not a full event log. Historical deadlocks, query plans and arbitrary calendar-window query history cannot be reconstructed from these tools alone. Tool limits and permissions can further restrict results.

    Yes. For example, the reviewed relationship tool ignores relationship_type, and its overview ignores schema_name. The autovacuum status tool uses a fixed threshold formula rather than actual global/per-table settings. Some tools with database_name select a connection but still query instance-wide views. The skill explains these scope differences, top-N limits, inconsistent dead-tuple ratio denominators and missing-feature placeholders rather than treating every tool description or heuristic label as a verified result.

    New in Data & Databases

    Free