Security brief

Historical advisory: prompt injection could reach LangChain SQL services

REVIEW

A separate 2023 LangChain advisory links prompt injection to execution against a connected SQL service.

Published 29 Oct 2026 · Material update 27 Oct 2026 — last time the underlying evidence changed · Evidence: high

What happened

GitHub’s reviewed advisory for CVE-2023-32785 describes a prompt-injection issue in LangChain versions before 0.0.247. Its stated impact is arbitrary code execution against the SQL service used by the chain.

The important boundary is between text processed by a model and instructions executed by a connected service. A model-generated command still needs the same authorization controls as a command created elsewhere.

This is historical coverage, not a newly discovered attack campaign. For legacy systems, compare the installed package with the advisory and review the permissions of the SQL connection. The advisory’s fixed version is a historical minimum; use a currently supported release after compatibility testing.

Read the original source.

Who should check?

Maintainers of older LangChain applications connected to SQL services.

Next step

Check CVE-2023-32785 and the connected database account’s permissions.

What we know

Affected package versions
Before 0.0.247

What remains unknown

Open question
The sources reviewed do not establish whether your deployment was exploited.

Source references

Technical details

CVE
CVE-2023-32785