---
id: "CVE-2026-120499"
kind: "vulnerability"
aliases: ["CVE-2026-120499"]
title: "h5io: crafted HDF5 file lets attackers run arbitrary Python code"
severity: "high"
cvss: 8.5
epss: null
exploited: false
status: "open"
category: "package"
products: []
ecosystems: []
tags: ["data-pipelines", "h5io", "hdf5", "package", "python"]
not_affected_if: ["trusted-files-only", "version-0.2.6-or-later"]
published_at: "2026-11-05T17:18:00+00:00"
first_seen_at: "2026-11-05T17:40:03+00:00"
updated_at: "2026-11-05T17:42:11+00:00"
sources: [{"source": "nvd", "url": "https://nvd.nist.gov/vuln/detail/CVE-2026-120499", "first_seen_at": "2026-11-05T17:40:03+00:00"}]
exploit_sources: []
---

# h5io: crafted HDF5 file lets attackers run arbitrary Python code

h5io is a Python library for reading and writing HDF5 files, and before version 0.2.6 its read_hdf5() function evaluates attacker-controlled content from a file. A crafted HDF5 file can make the reader run arbitrary Python code, through an eval() of a node named timezone, an arbitrary module import, or pickled data in pandas nodes.

_Summary, check and fix written by a model from the sources; labelled as such._

## How to check

- Check the installed h5io version with `pip show h5io` and compare it to the fixed release.
- Search your code for calls to `read_hdf5` and note whether it opens files from untrusted sources.

## Fix

- h5io: move to `0.2.6` or later.

## Timeline

- 2026-11-05T17:40:03+00:00 nvd https://nvd.nist.gov/vuln/detail/CVE-2026-120499

## Documents

- [advisory] Vendor advisory: github.com https://github.com/h5io/h5io/security/advisories/GHSA-vwvw-f6v4-w737
- [advisory] Third-party advisory: www.vulncheck.com https://www.vulncheck.com/advisories/h5io-before-0.2.6-arbitrary-code-execution-via-read-hdf5
- [fix] Patch: github.com https://github.com/h5io/h5io/commit/afff1d61b31790f8c27e6c577bb7703eb1883ddd
- [fix] Release notes: github.com https://github.com/h5io/h5io/releases/tag/h5io-0.2.6

## From the sources

h5io before 0.2.6 contains a code injection vulnerability in read_hdf5() that allows attackers to execute arbitrary code by supplying a crafted HDF5 file. Attackers can store Python code in a node titled timezone, which _triage_read() passes to eval() with builtins reachable, name an arbitrary module in class or custom-object nodes that _import_class() imports, or place pickled objects in pd_dataframe or pd_series nodes that pandas deserializes on read.

---

_Free to read, with fair use of about 60 pages a minute. For higher limits or API access: sales@infrastack.ai_