threatdeck
Live 14:45 UTC ·

AI Stack Alert Medium CVE-2026-125322 For CISO, AppSec, AI Eng, SOC / IR AI-assisted

CVE-2026-125322: pgEdge Postgres MCP Server count_rows tool allows cross-table data inference before 1.1.0

The where parameter of the count_rows tool in pgEdge Postgres MCP Server accepted subqueries against tables other than the one named in the tool call, prior to version 1.1.0. A caller could use predicates and the returned row count as an inference oracle to extract data from any table the connected database role could read. The fix is included in version 1.1.0.

Why it matters

Organizations exposing this MCP server to AI clients can have data from any table readable by the database role inferred by a caller. The database privilege boundary itself was not bypassed, so access is limited to what the role could already SELECT.

Do this

  1. Upgrade pgEdge Postgres MCP Server to version 1.1.0 or later.
  2. Review which database roles the MCP server connects with and remove SELECT privileges on tables that should not be reachable through AI clients.
  3. Audit MCP server logs for count_rows calls whose where parameter references tables other than the tool's target table.
  4. Restrict which MCP clients can reach the server until the upgrade is applied.

Exploit Clock

Exploitation unlikely (for now)

  • EPSS 0.33%

Signals

  • NoCISA KEV
  • 6.5CVSS
  • 0.33%EPSS
  • NoPublic PoC
  • 27/100Priority
  • 80%Confidence

Evidence

“Prior to 1.1.0, the where parameter of the count_rows tool accepted subqueries against tables other than the table named in the tool call.”

NVD (AI keyword search)

“The database privilege boundary remained intact, and the flaw did not grant access to tables outside that role's existing SELECT privileges.”

NVD (AI keyword search)

“This issue is fixed in version 1.1.0.”

NVD (AI keyword search)

Sources

Affected / related

pgEdge Postgres MCP Server, inference oracle via row count

Published 2026-11-10 18:17 UTC · updated 2026-11-12 16:46 UTC · AI-assisted summary, checked against quoted sources · How we score

Following

Follow the vendors, products, AI tools, techniques and actors you care about. Matching stories, CVEs and exploited bugs are marked, a Following tab appears in the feed, and returning visits show what changed. Saved only in this browser.

Presets

Get updates

A personal RSS feed of new stories on everything you follow. Paste it into any feed reader, Slack or Teams RSS app.