# Ray: Arbitrary code execution via ray.data.read_webdataset default decoder: pickle.loads(value) and torch.load(weights_only=False) Advisory record · `GHSA-hhrp-gw25-jr43` - **Identifiers:** `GHSA-hhrp-gw25-jr43` · `CVE-2026-57516` · `PYSEC-2026-2273` - **State:** Confirmed - **Severity:** CVSS 8.6 High - **Critical:** No - **Weakness (CWE):** [CWE-94](https://cwe.mitre.org/data/definitions/94.html) · [CWE-502](https://cwe.mitre.org/data/definitions/502.html) - **Tags:** [Code injection](https://agentprime.io/advisories/tag/code-injection.md) · [Insecure deserialization](https://agentprime.io/advisories/tag/deserialization.md) - **Published:** 24 Jul 2026 - **Updated:** 10 Sept 2026 - **Reviewed by a person:** No Published by the GitHub Advisory Database at 24 Jul 2026, 15:47 UTC. agentprime has not analysed this yet, so whether it affects any app is not known. ## Affected packages - `ray` (PyPI): affected `<2.56.0`; fixed in `2.56.0` ## Machine-written summary Summary written by Claude Haiku from the source advisory. Not reviewed by a person. > Ray versions before 2.56.0 are vulnerable to arbitrary code execution through ray.data.read_webdataset when its default decoder handles .pickle/.pkl and .pt/.pth members in a TAR file. An attacker who can supply a malicious TAR file that a user reads, from a shared bucket, an HTTP URL, or another source, can run code in the calling Ray process when the default decoder runs. Upgrade to Ray 2.56.0 or later. The source gives no workaround beyond avoiding untrusted TAR input or passing an explicit decoder. Written 4 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 - 8.6 High, CVSS 4.0: `CVSS:4.0/AV:N/AC:L/AT:N/PR:N/UI:A/VC:H/VI:H/VA:H/SC:N/SI:N/SA:N`, from CNA (VulnCheck) - **CISA KEV:** Not listed. - **EPSS:** 0.0086, the 58th percentile, on 10 Nov 2026: FIRST’s estimate of the chance of exploitation in the next 30 days. ## Timeline - 1 Jul 2026, 16:36 UTC: [Confirmed by the CVE List](https://www.cve.org/CVERecord?id=CVE-2026-57516) - 31 Oct 2026, 14:29 UTC: First seen by agentprime - 31 Oct 2026, 14:29 UTC: [Confirmed by OSV](https://osv.dev/vulnerability/GHSA-hhrp-gw25-jr43) ## Sources - [GitHub Advisory Database: GHSA-hhrp-gw25-jr43](https://github.com/advisories/GHSA-hhrp-gw25-jr43) - [OSV: GHSA-hhrp-gw25-jr43](https://osv.dev/vulnerability/GHSA-hhrp-gw25-jr43) - - - - - - - - - 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 (), 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: - In a browser: reads your lockfile in the browser, never uploading it. - Whether it touches your code, with file and line: the Claude Code plugin, - As JSON: , or `POST https://api.agentprime.io/v1/match` with your packages () --- The web page: . Everything for agents: .