# Picklescan (scan_pytorch) Bypass via dynamic eval MAGIC_NUMBER Advisory record · `GHSA-97f8-7cmv-76j2` - **Identifiers:** `GHSA-97f8-7cmv-76j2` · `CVE-2026-53875` · `PYSEC-2026-2873` - **State:** Confirmed - **Severity:** CVSS 7.1 High - **Critical:** No - **Weakness (CWE):** [CWE-95](https://cwe.mitre.org/data/definitions/95.html) · [CWE-184](https://cwe.mitre.org/data/definitions/184.html) - **Tags:** [Code injection](https://agentprime.io/advisories/tag/code-injection.md) · [CWE-184](https://agentprime.io/advisories/tag/cwe-184.md) - **Published:** 18 Feb 2026 - **Updated:** 13 Jul 2026 - **Reviewed by a person:** No Published by the GitHub Advisory Database at 18 Feb 2026, 17:45 UTC. agentprime has not analysed this yet, so whether it affects any app is not known. ## Affected packages - `picklescan` (PyPI): affected `<1.0.3`; fixed in `1.0.3` ## Machine-written summary Summary written by Claude Haiku from the source advisory. Not reviewed by a person. > The picklescan package before 1.0.3 can be bypassed when its scan_pytorch function checks a PyTorch file. A crafted PyTorch file can hide its magic number from the scanner's opcode check through a dynamic eval call in a reduce hook, so the scanner misses it while PyTorch still loads the value. An attacker who can get a victim to load such a file may achieve arbitrary code execution. Upgrade picklescan to version 1.0.3 or later. The source gives few details on the exact crafting steps. Written 6 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 - 7.1 High, CVSS 4.0: `CVSS:4.0/AV:N/AC:L/AT:N/PR:N/UI:P/VC:N/VI:H/VA:N/SC:N/SI:N/SA:N`, from CNA (VulnCheck) - **CISA KEV:** Not listed. - **EPSS:** 0.0055, the 46th percentile, on 7 Nov 2026: FIRST’s estimate of the chance of exploitation in the next 30 days. ## Timeline - 31 Oct 2026, 15:01 UTC: First seen by agentprime - 31 Oct 2026, 15:01 UTC: [Confirmed by OSV](https://osv.dev/vulnerability/GHSA-97f8-7cmv-76j2) ## Sources - [GitHub Advisory Database: GHSA-97f8-7cmv-76j2](https://github.com/advisories/GHSA-97f8-7cmv-76j2) - [OSV: GHSA-97f8-7cmv-76j2](https://osv.dev/vulnerability/GHSA-97f8-7cmv-76j2) - - - - - - - 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: .