On this page (7 sections)

A critical authorization bypass, CVE-2026-130773, affects Incus versions before 7.6.0. According to the NVD record, a user who is meant to be confined to a restricted project can instead create resources in the unrestricted default project. From there, the record describes two outcomes: running scripts as root, or reading host files.

If you run Incus as a shared platform, with several teams or tenants each limited to their own restricted project, this flaw concerns you. Hosting administrators and anyone operating multi-user Incus servers should read on.

What happened

Incus organizes instances, networks, volumes and other objects into projects. Projects can be marked as restricted, which limits what a user with access to that project is allowed to do. The NVD record for CVE-2026-130773 says this boundary can be bypassed: a restricted project user can target projects that lack features.* flags and, by doing so, create resources in the unrestricted default project instead.

The record was published on 2026-11-18. At the time of writing, the only source in our research is the NVD entry, which points to a GitHub security advisory from the Incus project. We have not independently verified anything beyond what those records state, and no other outlet has covered the story yet.

Technical details

The weakness class is CWE-863, incorrect authorization. In plain terms, a permission check exists but does not cover every path to the same result. Here, the NVD description says a request aimed at a project lacking features.* flags results in resources being created in the default project, which is not restricted.

The NVD description names two consequences:

  • Bridge networks: creating a bridge network with raw.dnsmasq dhcp-script injection, which the record says results in scripts running as root.
  • Custom storage volumes: creating a custom volume with block.create_options mkfs arguments, which the record says allows reading files from the host.

We are deliberately not describing how to craft such requests. As our own analysis rather than a statement from the record: both outcomes involve configuration keys that look like administrative settings, so it is sensible to treat any appearance of them in the default project as worth reviewing.

The NVD text does not state whether authentication is required beyond having restricted project access, nor whether remote exposure of the Incus API is a precondition. The description refers to restricted project users, so treat any account with access to a restricted project as the relevant threat model.

Who is affected

You are realistically exposed if all of the following are true:

  • You run Incus at a version earlier than 7.6.0.
  • You use projects to separate users or tenants, and some of them are restricted.
  • At least one person or automation you do not fully trust holds credentials scoped to a restricted project.

A single-admin lab where only you hold full access has lower practical risk, because there is no restricted user to abuse the gap. In a shared environment, the record's described outcome of root script execution is serious by nature, so treat it as a priority.

How to check your exposure

Start with the version. This command prints the client and server version of Incus; you are looking for a server version lower than 7.6.0. Run it on the Incus host.

Show Incus client and server version (run on the Incus host)Shell
incus version

Next, list your projects to see which ones exist and how they are configured. Run this on the host as a user with access to the server.

List Incus projects (read-only)Shell
incus project list

To see the settings of a specific project, including whether it is restricted and which features.* keys are set, show it by name. Replace the placeholder with a real project name. Look for projects that are restricted but do not have the features.* keys set the way you expect.

Show one project's configuration (read-only)Shell
incus project show PROJECT_NAME

Then review what already exists in the default project, since that is where the record says the unauthorized resources would be created. Look for networks and volumes you do not recognize, particularly any with raw.dnsmasq or block.create_options set.

List networks and storage volumes in the default project (read-only)Shell
incus network list --project default
incus storage volume list --project default

If you find an unfamiliar network, inspect its configuration with the command below, replacing the placeholder with the network name.

Show a network's configuration (read-only)Shell
incus network show NETWORK_NAME --project default
Tip. Compare the list of networks and volumes against your infrastructure records or configuration management. Anything in the default project that nobody on your admin team created deserves a closer look.

What to do now

  1. Confirm your Incus server version with incus version and note every host running a release earlier than 7.6.0.
  2. Upgrade to Incus 7.6.0 or later using your distribution's or vendor's normal package channel, and read the Incus advisory for any additional guidance.
  3. Review who holds credentials or trust tokens for restricted projects, and remove access that is no longer needed.
  4. Audit the default project for networks and volumes you did not create, especially any using raw.dnsmasq or block.create_options.
  5. If you find unexplained resources or suspect misuse, treat the host as potentially compromised: rotate secrets and credentials that were reachable from it and follow your incident process.
  6. Limit network exposure of the Incus API to trusted management networks while you work through the upgrade.
Caution. Upgrading Incus can restart the daemon and may interrupt management operations. Plan the change for a maintenance window and check that your instances come back as expected. Package names and upgrade commands differ by distribution and installation method, so use the procedure documented for how you installed Incus.

There is no workaround described in the research we have. Until you can upgrade, the most useful interim measure is reducing who can authenticate to restricted projects at all.

Detection and monitoring

Even after patching, monitoring for these configuration keys is a reasonable control. Periodically list networks and volumes in the default project and alert on new entries, and review the Incus server logs for unexpected create operations against the default project made by restricted users. Exact log locations and event formats depend on your installation, so confirm them in your own environment rather than assuming a path.

Management dashboards or reverse proxies in front of Incus should be kept off the public internet where possible. If you want outside-in visibility into what your servers expose, a Webforta scan of verified hosts can show what is reachable, and monitoring can alert you to certificate and DNS changes on management endpoints. Neither replaces patching, and neither detects this specific flaw.

What we do not know yet

  • No exploitation in the wild is reported in the research provided. That is an absence of information, not proof that nobody is exploiting it.
  • The NVD text does not include a proof of concept, and we are not providing one.
  • Only one independent publisher, the NVD, covers this so far. Details such as the exact fix and any additional affected configurations should be taken from the project's advisory as it is updated.

We will revisit this article if confirmed exploitation or further technical detail appears.

Sources

This article is based on the public sources below. Details can change as vendors and researchers publish more; check the original advisories before acting on version numbers.

  1. CVE-2026-130773: Incus before 7.6.0 contains an authorization bypass that allows restricted project users to create resources in the unrestricted default project by targeting…nvd.nist.gov ·
  • Incus
  • CVE-2026-130773
  • authorization bypass
  • containers
  • virtualization