H5io (< 0.2.6) - CraftedSignal Threat Feedhttps://feed.craftedsignal.io/products/h5io--0.2.6/Trending threats, MITRE ATT&CK coverage, and detection metadata. Fed continuously.Hugoenhello@craftedsignal.iohello@craftedsignal.ioThu, 05 Nov 2026 18:23:14 +0000https://feed.craftedsignal.io/favicon-32x32.pngCraftedSignal Threat Feedhttps://feed.craftedsignal.io/3232https://feed.craftedsignal.io/favicon.svgh5io: Arbitrary code execution in read_hdf5 (CVE-2026-120499)https://feed.craftedsignal.io/briefs/2026-11-h5io-rce/Thu, 05 Nov 2026 18:23:14 +0000hello@craftedsignal.iohttps://feed.craftedsignal.io/briefs/2026-11-h5io-rce/The h5io library before version 0.2.6 is vulnerable to arbitrary code execution via crafted HDF5 files that trigger unsafe eval, arbitrary imports, or insecure pickle deserialization during data loading.The h5io Python library, specifically in versions prior to 0.2.6, contains a code injection vulnerability within the read_hdf5() function. An attacker can craft a malicious HDF5 file containing specific node structures that, when processed, lead to arbitrary code execution. The vulnerability stems from three primary mechanisms:

  1. The timezone node is passed to eval() with builtins enabled.
  2. class or custom-object nodes allow for arbitrary module imports via _import_class().
  3. pd_dataframe or pd_series nodes contain pickled objects that are insecurely deserialized by pandas.

Impact

Successful exploitation allows for arbitrary code execution with the privileges of the application process that invokes the read_hdf5() function.

Recommendation

  • Upgrade the h5io library to version 0.2.6 or later immediately.
  • Audit applications that process externally sourced HDF5 files using h5io.
  • Implement sandboxing or containerization for untrusted file processing workloads to limit the impact of code execution vulnerabilities.
]]>highadvisory