---
id: "CVE-2026-87911"
kind: "vulnerability"
aliases: ["CVE-2026-87911"]
title: "AWS Postgres MCP server: read-only bypass allows OS commands on host"
severity: "critical"
cvss: 9.6
epss: null
exploited: false
status: "open"
category: "ai-tooling"
products: []
ecosystems: []
tags: ["ai-tooling", "aws", "lab-disclosure", "mcp", "postgres", "python", "self-hosted"]
not_affected_if: ["least-privilege-db-role", "managed-postgres-profile", "version-1.1.7-or-later"]
published_at: "2026-09-09T00:00:00+00:00"
first_seen_at: "2026-10-24T20:05:36+00:00"
updated_at: "2026-10-25T05:29:13+00:00"
sources: [{"source": "lab", "url": "https://aws.amazon.com/security/security-bulletins/rss/2026-104-aws/", "first_seen_at": "2026-10-24T20:05:36+00:00"}, {"source": "nvd", "url": "https://nvd.nist.gov/vuln/detail/CVE-2026-87911", "first_seen_at": "2026-10-24T20:15:27+00:00"}]
---

# AWS Postgres MCP server: read-only bypass allows OS commands on host

The awslabs postgres-mcp-server Python package before 1.1.7 fails to block COPY ... TO PROGRAM statements in its read-only SQL validation. Crafted content processed by an authenticated user of the MCP server can make the PostgreSQL host run operating system commands.

_Summary, check and fix written by a model from the sources; labelled as such._

## How to check

- Check the installed version with `pip show awslabs.postgres-mcp-server` and confirm it is `1.1.7` or later.
- Check whether the server uses the self-managed PostgreSQL profile (PG_WIRE_PROTOCOL connection method).
- Check whether the database role the server connects with is a superuser or holds pg_execute_server_program.

## Fix

- Upgrade awslabs.postgres-mcp-server to `1.1.7` or later from PyPI.
- Connect the MCP server with a dedicated least-privilege Postgres role, not a superuser, rds_superuser or the master user.
- Remove pg_execute_server_program and superuser from the role the server uses.

## Timeline

- 2026-10-24T20:05:36+00:00 lab https://aws.amazon.com/security/security-bulletins/rss/2026-104-aws/ — AWS: CVE-2026-87911 - Read-only enforcement bypass enabling operating system command execution in the SQL validation component of Amazon awslabs postgres-mcp-server
- 2026-10-24T20:15:27+00:00 nvd https://nvd.nist.gov/vuln/detail/CVE-2026-87911

## Documents

- [advisory] AWS: CVE-2026-87911 - Read-only enforcement bypass enabling operating system command execution in the SQL validation component of Amazon awslabs postgres-mcp-server https://aws.amazon.com/security/security-bulletins/rss/2026-104-aws/

## From the sources

CVE-2026-87911 - Read-only enforcement bypass enabling operating system command execution in the SQL validation component of Amazon awslabs postgres-mcp-server

### CVE-2026-87911 - Read-only enforcement bypass enabling operating system command execution in the SQL validation component of Amazon awslabs postgres-mcp-server

Bulletin ID: 2026-104-AWS Scope: AWS Content Type: Important (requires attention) Publication Date: 09/09/2026 12:30 PM PDT

Description:

awslabs.postgres-mcp-server is a python package that implements a Postgres MCP server. We found CVE-2026-87911, where an OS command injection weakness in the read-only enforcement of the SQL validation component in Amazon awslabs postgres-mcp-server before 1.1.7 might allow an unauthenticated actor to execute operating system commands on the host of a self-managed PostgreSQL server by placing a crafted COPY ... TO PROGRAM statement into content that is processed when an authenticated user interacts with the MCP server in its default read-only mode.

Impacted versions:

awslabs.postgres-mcp-server before 1.1.7, self-managed PostgreSQL deployment profile (PG_WIRE_PROTOCOL connection method) where the configured database role holds superuser or pg_execute_server_program.

Resolution:

This issue has been addressed in awslabs postgres-mcp-server (python package) version 1.1.7 in pypi.

Workarounds:

The strongest control is to connect the MCP server using a dedicated Postgres role that has only the privileges it actually needs, so that the database itself enforces the boundary regardless of what SQL reaches it. In particular:

- Do not connect as a superuser, rds_superuser, or the cluster master user. Those roles bypass row-level security, can read credential catalogs (pg_authid, pg_user_mappings), and can terminate other sessions.
- For read-only use, grant only CONNECT + USAGE + SELECT on the schemas the agent needs, and force read-only transactions at the role level.
- For read/write use, grant only the specific INSERT/UPDATE/DELETE privileges required, scoped to the necessary schemas and tables.

Combining a minimal-privilege role (database-enforced) with the blocklist (application-enforced) gives you defense in depth: even if a query slips past the blocklist, the role's privileges still bound what it can do.

References:

- CVE-2026-87911
- GHSA-fph8-pg5w-78fv

Acknowledgement:

We would like to thank Mordehai Attia (Corsen AI) and Ryan von Brubeck for collaborating on this issue through the coordinated vulnerability disclosure process.

 Please email aws-security@amazon.com with any security questions or concerns.

---

_Free to read at a human pace. An API with keys and an MCP server are in preparation; early access: hello@infrastack.ai_