# containerd CRI checkpoint restore CDI annotation smuggling

Advisory record · `GHSA-33vj-92qq-66hc`

- **Identifiers:** `GHSA-33vj-92qq-66hc` · `CVE-2026-53492` · `GO-2026-5064`
- **State:** Confirmed
- **Severity:** CVSS 8.4 High
- **Critical:** No
- **Weakness (CWE):** [CWE-20](https://cwe.mitre.org/data/definitions/20.html) · [CWE-863](https://cwe.mitre.org/data/definitions/863.html)
- **Tags:** [Broken access control](https://agentprime.io/advisories/tag/access-control.md) · [CWE-20](https://agentprime.io/advisories/tag/cwe-20.md)
- **Published:** 19 Jun 2026
- **Updated:** 10 Sept 2026
- **Reviewed by a person:** No

Published by the GitHub Advisory Database at 19 Jun 2026, 19:35 UTC. agentprime has not analysed this yet, so whether it affects any app is not known.

## Affected packages

- `github.com/containerd/containerd/v2` (Go): affected `>=2.1.0, <2.1.9 || >=2.2.0, <2.2.5 || >=2.3.0, <2.3.2`; fixed in `2.1.9`, `2.2.5`, `2.3.2`

## Machine-written summary

Summary written by Claude Haiku from the source advisory. Not reviewed by a person.

> containerd (Go module github.com/containerd/containerd/v2) versions 2.1.0 before 2.1.9, 2.2.0 before 2.2.5, and 2.3.0 before 2.3.2 are affected. When restoring a container from a checkpoint, containerd keeps CDI annotations from untrusted checkpoint metadata, so a user with pod creation permissions can inject arbitrary CDI edits such as device nodes and host mounts. Exploitation requires CDI to be enabled on the node with a matching host CDI specification. Upgrade to containerd 2.1.9, 2.2.5, or 2.3.2; containers restored from untrusted checkpoints may need to be recreated.

Written 1 Nov 2026.

## Not analysed yet

We haven’t published our analysis of this Advisory yet. Once a person has reviewed it, this page will say what’s vulnerable, what has to be true for your app to be affected, and how to check.

## Severity and exploitation

- 8.4 High, CVSS 4.0: `CVSS:4.0/AV:N/AC:L/AT:N/PR:L/UI:N/VC:N/VI:H/VA:N/SC:H/SI:H/SA:N`, from CNA (GitHub_M)

- **CISA KEV:** Not listed.
- **EPSS:** 0.0035, the 29th percentile, on 18 Nov 2026: FIRST’s estimate of the chance of exploitation in the next 30 days.

## Timeline

- 18 Jun 2026, 23:23 UTC: [Published by the maintainers as a GitHub repository advisory](https://github.com/containerd/containerd/security/advisories/GHSA-33vj-92qq-66hc)
- 31 Oct 2026, 14:13 UTC: First seen by agentprime
- 31 Oct 2026, 14:13 UTC: [Confirmed by OSV](https://osv.dev/vulnerability/GHSA-33vj-92qq-66hc)

## Sources

- [GitHub Advisory Database: GHSA-33vj-92qq-66hc](https://github.com/advisories/GHSA-33vj-92qq-66hc)
- [OSV: GHSA-33vj-92qq-66hc](https://osv.dev/vulnerability/GHSA-33vj-92qq-66hc)
- <https://github.com/containerd/containerd/security/advisories/GHSA-33vj-92qq-66hc>
- <https://github.com/containerd/containerd>

Facts on this page come from the GitHub Advisory Database (CC-BY 4.0), OSV and the other sources listed, as published. We don’t reproduce full advisory descriptions, which can contain exploit code.

## Check whether this affects you

A Match means an installed version is in the affected range above. Most don’t affect the app; agentprime hasn’t analysed this one yet.

- With the agentprime MCP server (<https://api.agentprime.io/mcp>), ask your agent to check the project: `check_lockfile` sends a lockfile, `check_packages` a list of installed packages, and both return the Matches and the upgrades that clear them. `get_advisory` returns this record. Setup: <https://agentprime.io/mcp.md>
- In a browser: <https://agentprime.io/check/> reads your lockfile in the browser, never uploading it.
- Whether it touches your code, with file and line: the Claude Code plugin, <https://agentprime.io/plugin.md>
- As JSON: <https://api.agentprime.io/v1/advisories/GHSA-33vj-92qq-66hc>, or `POST https://api.agentprime.io/v1/match` with your packages (<https://agentprime.io/api.md>)

---

The web page: <https://agentprime.io/advisories/GHSA-33vj-92qq-66hc/>. Everything for agents: <https://agentprime.io/llms.txt>.
