AI Sec Watch: A Security Intelligence Platform for AI Systems

Luu, T.J.

LowVulnerability

GHSA-5x6v-p487-7qh2: LangChain: RediSearch Filter Injection via Unescaped Tag/Text Values

Source: GitHub Advisory(opens in a new tab)
Identifiers: CVE-2026-105799GHSA-5x6v-p487-7qh2
Published: October 6, 2026
Record updated: October 9, 2026

View JSON

Affected: @langchain/redis <= 1.1.0
Fixed in: 1.1.1
Known exploitation: Not listed in the CISA Known Exploited Vulnerabilities catalog at the last check.
EPSS: 0.3%

Summary

`@langchain/redis` versions through 1.1.0 did not properly escape values used to build structured RediSearch TAG and TEXT filters. An attacker who controls values passed into these filters can inject RediSearch syntax, altering or broadening the search query and potentially exposing indexed documents outside their intended scope where the filter serves as a tenant or document-access boundary.

Mitigation

Upgrade to `@langchain/redis` 1.1.1 or later.

Details

Scores and exploitation

Patch available: Yes
EPSS: 0.3%Estimated probability that this CVE is exploited in the next 30 days, from the FIRST Exploit Prediction Scoring System. Read on October 9, 2026.
Disclosure date: October 6, 2026

Weaknesses and techniques

AI threat technique (MITRE ATLAS): AML.T0020
AML.T0051.001

Affected

Packages: @langchain/redis@<= 1.1.0 (fixed: 1.1.1)
Vendors: LangChain

Classification

Attack type: Other
Attack sophistication: Moderate
AI component targeted: RAG
Impact: Confidentiality,
Integrity
Labels: security

How this record was produced

Summaries, mitigations and labels on this page are generated automatically from the source. The linked source is authoritative.

Original source: https://github.com/advisories/GHSA-5x6v-p487-7qh2

First tracked: October 6, 2026 at 02:00 PM ET, from GitHub Advisory Database

Classifier: claude-haiku-5-5, prompt v4

Classifier confidence: 85% (self-assessed)

Summary

Mitigation

Related items