Threat reportVulnerabilityTL-2026-6357
Hostile Inference: Docker Model Runner Argument Injection and Container Escape Chains in Docker Desktop (CVE-2026-28400, ZDI-26-150/258/451)
ZDI research shows the unauthenticated Docker Model Runner (DMR) _configure API, reachable from any default container at model-runner.docker.internal, allows llama-server argument injection (CVE-2026-28400). Chained with host file read/write, a macOS Seatbelt sandbox escape (ZDI-26-451) and Docker Extensions abuse (ZDI-26-258), it enables container-to-host escape on Docker Desktop for macOS and Windows (WSL2). First disclosed .
Key facts
| Identifiers | TL-2026-6357, CVE-2026-28400 |
|---|---|
| Also known as | Hostile Inference, ZDI-26-150, ZDI-26-258, ZDI-26-451, GHSA-m456-c56c-hh5c |
| Severity | HIGH |
| CVSS | 7.5 (CVSS:3.1/AV:L/AC:H/PR:L/UI:R/S:C/C:H/I:H/A:H) |
| First disclosed | |
| EPSS | CVE-2026-28400: 0.2% (17.6th percentile) |
| Affected products | Docker Docker Model Runner (github.com/docker/model-runner); Docker Docker Desktop (macOS and Windows/WSL2) |
| Weaknesses (CWE) | CWE-749 |
| Tools | llama.cpp |
| MITRE ATT&CK techniques | T1005 Data from Local System, T1033 System Owner/User Discovery, T1059 Command and Scripting Interpreter, T1485 Data Destruction, T1565.001 Stored Data Manipulation, T1611 Escape to Host |
| Motivation | UNKNOWN |
| Target sectors | technology, software-development, artificial-intelligence |
| Target regions | Global |
| Detection rules | 9 (1 shown free) |
| Sources | 8 cited sources |
Detection coverage
As of 2026-11-18, Threadlinqs Intelligence publishes 9 detection rule(s) for TL-2026-6357 across Splunk SPL, Microsoft KQL and Sigma, covering 13 indicator(s) of compromise. The whole corpus is readable without an account. Every threat comes with one complete detection in Splunk SPL, Microsoft KQL and Sigma, free for everyone; a free account adds daily TL Creds to unlock the rest. Paid plans add a larger daily allowance, correlation and the MCP server. Threadlinqs MCP server · View plans.
Free detection: Docker Model Runner llama-server Argument Injection (CVE-2026-28400) — Exploitation
Detects the host-side llama.cpp llama-server process (child of the Docker Backend) started with attacker-injected runtime flags delivered through the unauthenticated DMR POST /engines/_configure API: --log-file (arbitrary host file write/overwrite, e.g. Docker.raw), --trust-remote-code (vLLM container-escape flag) or a direct Docker.raw reference. This is the core CVE-2026-28400 / ZDI-26-150 / GHSA-m456-c56c-hh5c exploitation artifact in TL-2026-6357.
- Languages: Splunk SPL, Microsoft KQL, Sigma
- ATT&CK: T1611, T1068, T1565.001, T1485, T1033
- Data sources: sysmon, windows_process_creation, mde_deviceprocessevents
- Severity: high
- Confidence: high
Splunk SPL
index=windows sourcetype="XmlWinEventLog:Microsoft-Windows-Sysmon/Operational" EventCode=1 (Image="*\\llama-server.exe" OR OriginalFileName="llama-server*") (CommandLine="*--log-file*" OR CommandLine="*--trust-remote-code*" OR CommandLine="*Docker.raw*")
| eval cmd_lc=lower(CommandLine)
| eval injected_flag=case(like(cmd_lc,"%--log-file%"),"--log-file", like(cmd_lc,"%--trust-remote-code%"),"--trust-remote-code", true(),"docker.raw")
| eval targets_vm_disk=if(like(cmd_lc,"%docker.raw%"),1,0)
| stats count min(_time) as first_seen max(_time) as last_seen values(CommandLine) as command_lines values(ParentImage) as parent_images by host, User, injected_flag, targets_vm_disk
| convert ctime(first_seen) ctime(last_seen)Microsoft KQL
DeviceProcessEvents
| where Timestamp > ago(24h)
| where FileName in~ ("llama-server", "llama-server.exe") or ProcessVersionInfoOriginalFileName startswith "llama-server"
| where ProcessCommandLine has_any ("--log-file", "--trust-remote-code", "Docker.raw")
| extend InjectedFlag = case(ProcessCommandLine has "--log-file", "--log-file", ProcessCommandLine has "--trust-remote-code", "--trust-remote-code", "Docker.raw")
| project Timestamp, DeviceName, AccountName, FileName, ProcessCommandLine, InitiatingProcessFileName, InitiatingProcessCommandLine, InjectedFlag
| order by Timestamp descSigma
title: Docker Model Runner llama-server Injected Runtime Flags (CVE-2026-28400)
id: 3f6b1c52-8a4e-4d7b-9c21-5e0a7b8d6f01
status: experimental
description: Detects llama-server (llama.cpp) spawned by Docker Model Runner with injected flags such as --log-file or --trust-remote-code, indicating abuse of the unauthenticated /engines/_configure API (CVE-2026-28400).
references:
- https://www.thezdi.com/blog/2026/11/17/hostile-inference-exploiting-docker-model-runner-for-container-escapes
- https://github.com/docker/model-runner/security/advisories/GHSA-m456-c56c-hh5c
- https://nvd.nist.gov/vuln/detail/CVE-2026-28400
author: Threadlinqs Team
date: 2026/11/18
tags:
- attack.privilege-escalation
- attack.t1611
- attack.t1068
- attack.impact
- attack.t1565.001
- attack.t1485
- cve.2026-28400
logsource:
category: process_creation
product: windows
detection:
selection_img:
Image|endswith: '\llama-server.exe'
selection_flags:
CommandLine|contains:
- '--log-file'
- '--trust-remote-code'
- 'Docker.raw'
condition: selection_img and selection_flags
falsepositives:
- Administrators intentionally running llama-server with --log-file for local debugging
- Developers testing llama.cpp outside Docker Model Runner
level: high
Known false positives:
- Developers or admins manually launching llama-server with --log-file for troubleshooting
- Local AI tooling that legitimately passes logging flags to llama-server
Tuning notes: Restrict to llama-server instances whose ParentImage is the Docker Desktop Backend to focus on DMR; allow-list known log paths under the Docker Desktop data directory; treat any --log-file value pointing at Docker.raw, a Startup folder, .ssh or another user-data path as critical. After upgrading to Docker Desktop 4.61.0+ (DMR 1.0.16) expect hits only from manual execution.
- Container Tooling Calling Docker Model Runner _configure API (CVE-2026-28400) · unlock with TL Creds
- llama-server Writing [indicator] or Sensitive Host Paths via --log-file Injection · unlock with TL Creds
- Docker Extension Installation Outside Marketplace Workflow (ZDI-26-258) · unlock with TL Creds
- HTTP POST to Docker Model Runner [indicator] Endpoints · unlock with TL Creds
- llama-server Process Referencing UNC Paths — Username Disclosure and Host File Read Probing · unlock with TL Creds
- macOS Seatbelt Sandbox Escape — llama-server or sandbox-exec Launching Apps via LaunchServices (ZDI-26-451) · unlock with TL Creds
- Docker Desktop backend.sock [indicator] Tampering — [indicator] Disabled (ZDI-26-258) · unlock with TL Creds
- llama-server Spawning Shells or LOLBins on Windows — DMR Host Code Execution · unlock with TL Creds
Indicators of compromise
As of 2026-11-18, Threadlinqs Intelligence tracks 13 indicators of compromise for TL-2026-6357.
Sample indicators (1 of 13)
- Domain:
model-runner[.]docker[.]internal
Defanged, TLP:CLEAR. Full set of 13 indicators: unlock with TL Creds.
Technical analysis
No in-the-wild exploitation is stated.
Docker Model Runner (DMR) is a Docker Desktop feature (enabled by default since 4.46.0) that runs llama.cpp's llama-server on the host, as a child of the Docker Backend process, to serve local models. Its REST API is unauthenticated and reachable from any container over HTTP at model-runner.docker.internal. The POST /engines/_configure and /engines/llama.cpp/_configure endpoints accepted runtime flags that were passed to llama-server without adequate validation (CWE-749, argument injection). Injecting the --log-file flag lets a container write or overwrite arbitrary files reachable by the Model Runner process, for example the Docker Desktop VM disk Docker.raw. ZDI-26-150 / CVE-2026-28400 was reported as a denial of service on macOS; ZDI's follow-up research (Nitesh Surana, Trend/TrendAI Research) extended the primitive toward host code execution under non-default configurations. The GitHub advisory GHSA-m456-c56c-hh5c rates it CVSS 7.5 (AV:L/AC:H/PR:L/UI:R/S:C).
Platform sandboxing differs. On macOS, llama-server runs under a Seatbelt profile (sandbox-exec, since Docker Desktop 4.46.0) that starts from (allow default) and denies selected operations, so some user-data paths under the Docker Desktop container directory remain reachable. On Windows, llama-server runs in a Job Object that limits memory, CPU, process count and lifetime but adds no filesystem restrictions beyond the user's own permissions. The post describes certain llama-server options performing host file reads and writes, leaking the Docker Desktop username through UNC-path handling and error output (needed to build user-specific paths), and sinks that put attacker-supplied metadata into inference logs for partial control over written content. Docker Extensions are described as the route to host command execution, since extension UIs can invoke privileged Docker CLI operations and unsandboxed host binaries.
Two related issues complete the chain. ZDI-26-451 (CVSS 8.8, macOS): the Seatbelt profile did not deny the launchservicesd Mach services, so a sandboxed process could launch applications through LaunchServices outside the sandbox (TCC and Gatekeeper still applied); fixed by commit 98038869 denying Mach lookups for com.apple.launchservicesd and com.apple.coreservices.launchservicesd. ZDI-26-258 (CVSS 8.2, Windows/WSL2 0-day): from a compromised Docker Desktop VM, the backend socket /run/host-services/backend.sock exposes an /app/settings API (GET/POST); changing onlyMarketPlaceExtensions re-enables installation of arbitrary extensions, linked to host code execution. Docker rejected this as outside its threat model because a container-to-VM escape is already required; no fix is described.
Remediation for CVE-2026-28400: DMR v1.0.16 (Docker Desktop 4.61.0+) removed then re-added runtime flags with path-safety validation, added a strict per-backend flag allowlist, and removed the allowlisted --trust-remote-code flag that could let vLLM models escape the container. Enhanced Container Isolation (ECI) blocks container access to Model Runner. The ZDI post and sources state no CVSS for the full chain, no in-the-wild exploitation, no attribution and no malicious IOCs; the indicators below are artifacts for hunting and detection rather than attacker infrastructure.
MITRE ATT&CK techniques by tactic:
Collection
Discovery
T1033 System Owner/User Discovery
Execution
T1059 Command and Scripting Interpreter
Impact
T1485 Data Destruction; T1565.001 Stored Data Manipulation
Privilege Escalation
Remediation
Patches
- Docker Model Runner v1.0.16 / Docker Desktop 4.61.0+ (GHSA-m456-c56c-hh5c)
- docker/model-runner commit 98038869: deny Mach lookups for launchservicesd services (ZDI-26-451)
Immediate actions
- Upgrade Docker Desktop to 4.61.0 or later (Docker Model Runner v1.0.16+) to fix CVE-2026-28400 / ZDI-26-150
- Enable Enhanced Container Isolation (ECI), which blocks container access to Model Runner
- Keep the marketplace-only extensions setting (onlyMarketPlaceExtensions) enabled and review installed Docker Extensions
Workarounds
- Enable Enhanced Container Isolation (ECI)
- Disable Docker Model Runner where not needed and do not expose it over localhost TCP
- ZDI-26-258: restrict interaction with the product; no vendor fix described
Longer-term hardening
- Treat containers, and mounted Docker sockets, as high-risk since they are prerequisites for the VM-to-host pivots
- Review local-AI tooling for unauthenticated local APIs and for sandboxes that do not restrict filesystem access
- Apply the latest Docker Desktop for macOS to receive the tightened Seatbelt profile (ZDI-26-451)
Affected and fixed versions
Timeline
- ZDI-26-258 (Docker Extensions VM-to-host pivot) reported to Docker (advisory lists 05/28/25 and 06/04/25 as submission dates)
- Docker states the ZDI-26-258 attack scenario is outside its security threat model
- ZDI-26-150 (DMR argument injection) reported to Docker
- Docker rejects the ZDI-26-258 report, stating exploitation requires prior privileged access
- DMR PR #482 (0699420a) removes runtime flags from POST /engines/_configure
- DMR PR #518 (41dd3f13) reintroduces runtime flags with validatePathSafety rejecting '/' and '\'
- DMR PR #601 (9acf1ed9) adds a strict per-backend flag allowlist
- ZDI-26-451 (macOS Seatbelt sandbox escape) reported to Docker
- GHSA-m456-c56c-hh5c / CVE-2026-28400 published (CVSS 7.5); tightened Seatbelt profile merged for ZDI-26-451
- ZDI-26-150 coordinated public release; fix shipped in DMR v1.0.16 / Docker Desktop 4.61.0
- ZDI publishes ZDI-26-258 as a 0-day advisory (CVSS 8.2); no fix described
- ZDI-26-451 coordinated public release (CVSS 8.8); Seatbelt fix commit 98038869
- ZDI blog 'Hostile Inference' publishes the full DMR container-escape research (URL path dated 2026/11/17)
Sources cited
- NVD: CVE-2026-28400 NVD (NIST) · Advisory · 2026-02-27
- ZDI-26-150: Docker Desktop for Mac Docker Model Runner zerodayinitiative.com · Advisory · 2026-03-03
- ZDI-26-258: (0Day) Docker Desktop extension-manager Exposed Dangerous Function Local Privilege Escalation zerodayinitiative.com · Advisory · 2026-04-15
- ZDI-26-451: Docker Desktop for macOS Inference Server Permissive Allow List Sandbox Escape zerodayinitiative.com · Advisory · 2026-07-23
- Hostile Inference: Exploiting Docker Model Runner for Container Escapes (Nitesh Surana, ZDI) thezdi.com · Research · 2026-11-18
- SentinelOne Vulnerability Database: CVE-2026-28400 SentinelOne · Research · 2026-03-02
- VulDB: Docker model-runner up to 1.0.15 API /engines/_configure vuldb.com · Research · 2026-03-02
- GHSA-m456-c56c-hh5c: Docker Model Runner Unauthenticated Runtime Flag Injection via _configure Endpoint GitHub · Code repository · 2026-02-27
Other reports on CVE-2026-28400
Drafted by the Threadlinqs research pipeline from 8 sources; published ; last updated —.