CVE-2026-123653 · AI agents & MCP
NO FIX YET · SEE WHAT TO DO
DeepSeek-Reasonix, an AI coding tool, fails to neutralize input in its discoverDocs function, so crafted content can inject instructions into the LLM prompt. The sources say an attack can be started remotely and that a public exploit has been disclosed.
first public Oct 27, 2026 · 1 source
Check it with your coding agent
It checks your repository and tells you if you’re affected, with the fix.
Teams running DeepSeek-Reasonix up to version 2.52.0 where the tool discovers and feeds documents into the model prompt. The sources do not say whether a default configuration is affected or what the document sources are, so whether a given setup is exposed is unknown.
Not affected if version-2.53.0-or-later
Who cares ai-coding-tools, deepseek-reasonix, llm-prompting
2.52.0 are affected.internal/memory/doc.go (discoverDocs) in your build.2.53.0, per the sources.A security vulnerability has been detected in esengine DeepSeek-Reasonix up to 2.52.0. Affected is the function discoverDocs of the file internal/memory/doc.go. The manipulation leads to improper neutralization of input used for llm prompting. The attack can be initiated remotely. The exploit has been disclosed publicly and may be used. Upgrading to version 2.53.0 is able to address this issue. The identifier of the patch is 4ca33493a673c2788aa8598fbca2b2f47a69e1b8. It is suggested to upgrade the affected component.
This page as .md ·
Are you affected?
Tell InfraStack what you run. It checks this one, and every next one, against your stack.
Ask from here. InfraStack checks the affected versions against your stack and answers in your inbox.
InfraStack checks the affected versions against your stack and answers in Slack, with the evidence and the fix.
NVD data is provided by NIST and is public domain. Posts quoted belong to their authors.