Docker Model Runner container-to-host code execution via unsandboxed trust_remote_code in Python inference backends
Title source: cnaExploitation Summary
EIP tracks 1 public exploit for CVE-2026-5817. PoCs published by gouldnicholas.
AI-analyzed exploit summary This repository contains a functional proof-of-concept exploit for CVE-2026-5817, demonstrating a remote code execution (RCE) vulnerability in Docker Model Runner. The exploit leverages a malicious OCI registry to serve a crafted model with an embedded Python payload, exploiting the lack of digest verification and the default trust_remote_code behavior in Python backends.
Description
The vllm-metal inference backend in Docker Model Runner on macOS unconditionally sets trust_remote_code=True when loading model tokenizers, and runs without sandboxing. This causes transformers.AutoTokenizer.from_pretrained() to import and execute arbitrary Python files included in any model pulled from an OCI registry, resulting in arbitrary code execution on the Docker host as the Docker Desktop user when inference is triggered. Any container on the Docker network can trigger this by calling the model-runner.docker.internal API to pull a malicious model and request inference.
Exploits (1)
This repository contains a functional proof-of-concept exploit for CVE-2026-5817, demonstrating a remote code execution (RCE) vulnerability in Docker Model Runner. The exploit leverages a malicious OCI registry to serve a crafted model with an embedded Python payload, exploiting the lack of digest verification and the default trust_remote_code behavior in Python backends.
References (1)
Scores
CVSS:3.1/AV:L/AC:L/PR:L/UI:R/S:C/C:H/I:H/A:H