AAK-MCP-LANGCHAIN-REDIS-CVE-2026-105799-001: LangChain.js Redis vector store below 1.1.1: unescaped filter values widen a RediSearch query
- Severity
- LOW
- Category
- Supply Chain (
supply-chain) - Scanner
agent_audit_kit/scanners/mcp_cve_pins_2026_07.py- Auto-fixable
- No.
- SARIF name
LangChainRedisFilterInjection
What it detects
@langchain/redis is LangChain.js's Redis vector store. Before 1.1.1 it put filter values into RediSearch queries unescaped, in the structured TAG and TEXT filters, and only partly escaped in the older string-array filter, so injected RediSearch syntax could change or widen the query. An application that builds a tenant or document-access boundary from a filter an attacker can influence can then return indexed documents outside that boundary (CVE-2026-105799, CVSS 4.0 2.3, CWE-943, GHSA-5x6v-p487-7qh2). 1.1.1 escapes every filter value (langchainjs#10701).
How to fix
Upgrade @langchain/redis to 1.1.1 or later and pin it. Until then, do not build a tenant or access boundary from a filter value a user controls, or enforce the boundary outside the vector search.
To confirm the finding is gone, report only this rule:
agent-audit-kit scan . --rules AAK-MCP-LANGCHAIN-REDIS-CVE-2026-105799-001
Limitations
Reports @langchain/redis below 1.1.1 where a package.json, lockfile or MCP config states the version, taking the lowest version a range allows. latest and other tags state no version and are not reported. It reads the version, not the filters, so a project whose filters never carry user input is reported without being exposed. The Python langchain-redis is a different package and is not read. A file over 2 MB, or a lockfile over 32 MiB, is not read. A declaration nested more than 64 levels deep in package.json, a TOML manifest or an MCP config (an MCP server more than 62) is still judged, and reported without a line; a lockfile entry keeps its line at any depth.
References
- CVE: CVE-2026-105799
- OWASP Agentic Top 10: ASI04 (Agentic Supply Chain Vulnerabilities), ASI06 (Memory & Context Poisoning)
- Adversa AI MCP Top 25 (AAK-defined ids): ADV-SUPPLY-04 (Known Vulnerable Dep)
- CSA AI Controls Matrix (AICM): STA-08, AIS-07, DSP-04
All rules. This page is generated from the rule registry when the documentation is built; to change it, change the rule in agent_audit_kit/rules/builtin.py. To give the rule a hand-written write-up, add docs/rules/AAK-MCP-LANGCHAIN-REDIS-CVE-2026-105799-001.md: the page then shows that text, with the registry reference appended under it.