LM Studio CVE-2026-119979: A Crafted Model Repo Could Plant a Payload That Runs on Every Launch

VulnCheck published CVE-2026-119979 on October 30, 2026: LM Studio versions before 0.4.22+1 allowed a crafted model repository to write files outside the models directory and plant configuration that ran an attacker’s Python payload on every launch. The finder is Ariel Fogel of Pillar Security.

Christopher Ort
By Christopher Ort · 2 min read
Share
A home office shelf of a desktop computer and an external drive, a model-download window out of focus on the monitor, evening light.
On this page
  1. Quick Take
  2. What happened
  3. What you should do
  4. What it means
  5. About the analysis
  6. Sources

Downloading a model should not be the same as installing software. The Injection reports that VulnCheck, on October 30, 2026, published CVE-2026-119979 (CVSS 8.7) against LM Studio: versions before 0.4.22+1 “let a crafted model repository write files outside the models directory, planting configuration that runs an attacker’s Python payload on every launch.” Ariel Fogel of Pillar Security found the issue.

i10X Quick Take

Quick Take

  1. Product: LM Studio, a popular local-model desktop app.
  2. Bug: A malicious model repository could write outside the models folder and install a payload that executed on each subsequent launch (CVE-2026-119979, CVSS 8.7).
  3. Fix: Update LM Studio to 0.4.22+1 or later.
  4. If you imported models from untrusted repos before the patch: do not just update; inspect or reset application config and consider a clean reinstall.

What happened

LM Studio users often pull models from shared repositories and community sources. If the import path trusts filenames or archive entries from that repository, a crafted package can write into configuration or startup locations instead of the models directory. Once a Python payload is planted to run at launch, every later start of LM Studio is code execution, even if the user never “opens” the malicious model again.

That is a supply-chain bug in the costume of a model download. It is also the kind of thing a prompt-injected agent, or a poisoned README, could steer a user toward.

What you should do

  • Update to LM Studio 0.4.22+1 or later on every machine that runs it.
  • If you downloaded models from unofficial repositories before updating, treat the app config as untrusted: look for unexpected startup scripts or Python files outside the models folder, or reinstall cleanly and re-import models from sources you trust.
  • Do not import model archives from strangers, even after the patch; the update closes this write path, not every future packing trick.

What it means

Local-model apps have become everyday developer tools, and they inherit the same unpack-and-trust problems that browser extensions and IDE plugins had for years. A CVSS 8.7 “runs on every launch” bug is not a theoretical sandbox debate; it is persistence on a laptop that also holds source code and cloud keys. Patch, then check what you already imported.

About the analysis

Researched and written with AI assistance under the editorial responsibility of the i10X News team. The title image is an AI-generated illustration. Read our AI Use Policy. Spotted an error? Request a correction.

This i10x report is based on The Injection’s week-44 summary of the VulnCheck advisory for CVE-2026-119979 (October 30, 2026), credited to Ariel Fogel of Pillar Security. i10x did not test LM Studio.

Sources

Christopher Ort

Christopher Ort

Christopher Ort handles SEO, AI agent workflows and PR & partnerships at i10x.ai, the all-in-one AI workspace built by i10X Global Pte. Ltd. in Singapore. He is a graduate of Friedrich-Alexander-Universität Erlangen-Nürnberg (FAU). He is the author of “The Style Penalty”, a study published on SSRN that analyses 1,576 evaluation decisions by four frontier AI models to show how AI-written CVs fare in automated hiring screens; the research has been covered by heise online and t3n.

Editor's picks

  1. CVE-2026-77604: Path Traversal Vulnerability in Code Graph RAG MCP Server Allows Arbitrary File Read
  2. CVE-2026-123271: ChatCLI OAuth Flaw Allows Token Theft via Malicious MCP Servers
  3. CVE-2026-123022: AgentScope MCP Workspace Allows Unauthenticated Remote Code Execution
  4. CVE-2026-97677: Path Traversal Flaw in IBM Langflow OSS 1.0.0–1.12.2 Allows File Read/Write
  5. Active Support DoS Flaw Lets Unauthenticated Attackers Crash Rails Servers via Action Cable