CVE-2026-6587: vibrantlabsai RAGAS SSRF in Collections Module

CVE-2026-6587 · Severity: medium · CVSS 6.3 · Published 2026-04-20

A security vulnerability exists in RAGAS, a framework used to evaluate Retrieval-Augmented Generation (RAG) systems. By providing specially crafted input, an attacker can force the server to read local files or access internal network services that are not intended to be public. This could lead to the theft of sensitive configuration files, cloud credentials, or internal data, potentially compromising the entire server infrastructure.

Vendors: PyPI.

Quick answers

Is CVE-2026-6587 exploited in the wild?
No. Junglewise Threat Intelligence has no report of CVE-2026-6587 being exploited in the wild.
How severe is CVE-2026-6587?
Medium severity, with a CVSS score of 6.3, published 20 April 2026.
Which products and versions are affected?
ragas (PyPI) >= 0.2.3, <= 0.4.3, ECOSYSTEM:introduced=0.2.3,last_affected=0.4.3.
Is there a fix for CVE-2026-6587?
No fixed version is recorded yet. Check the vendor's advisory in the references below. If you run it, Junglewise drafts a mitigation that fits your setup and applies it once you approve, with senior security engineers behind it.
How is CVE-2026-6587 attacked?
It is attacked over the network, in a software library.
Am I affected by CVE-2026-6587?
Check whether you run ragas (PyPI) at an affected version. Junglewise does that check for you around the clock: connect your code, cloud and domains, and it matches new vulnerabilities like this one against what you actually run and tells you when one reaches you.

Technical details

A Server-Side Request Forgery (SSRF) vulnerability exists in the vibrantlabsai RAGAS library up to version 0.4.3. The flaw is located in the _try_process_local_file and _try_process_url functions within src/ragas/metrics/collections/multi_modal_faithfulness/util.py. An attacker can manipulate the 'retrieved_contexts' argument to bypass image validation checks using URI schemes like 'file://' combined with URL fragments (e.g., #payload.jpg) to trick the mimetypes detection. This allows for arbitrary local file reads and internal network probing via urllib.request.urlopen. While a similar issue (CVE-2025-45691) was previously patched in a different module, this specific component remains vulnerable. No official patch from the vendor is currently confirmed.

Affected products

  • vibrantlabsai RAGAS up to 0.4.3
  • Red Hat OpenShift AI (RHOAI)

Identifiers

  • CVE-2026-6587
  • GHSA-95ww-475f-pr4f
  • PYSEC-2026-3046

Timeline

  • 2026-04-20: disclosed: Initial public disclosure of the vulnerability.
  • 2026-04-20: advisory: NVD and Red Hat published advisories.

References

This page as Markdown. Free to reuse under CC BY 4.0 with a link back.