CVE-2026-46526: LearningCircuit Local Deep Research SSRF via parser differential

CVE-2026-46526 · Severity: medium · CVSS 5 · Published 2026-05-28

Local Deep Research is an AI-powered research assistant used for conducting iterative, deep-dive research. A security flaw in how the application validates web addresses (URLs) allows attackers to bypass safety checks and force the system to make requests to internal servers or cloud metadata services. This could lead to the exposure of sensitive internal data or cloud credentials, potentially compromising the infrastructure where the assistant is hosted.

Technologies: local-deep-research (PyPI). Vendors: PyPI.

Quick answers

Is CVE-2026-46526 exploited in the wild?
No. Junglewise Threat Intelligence has no report of CVE-2026-46526 being exploited in the wild.
How severe is CVE-2026-46526?
Medium severity, with a CVSS score of 5, published 28 May 2026.
Which products and versions are affected?
local-deep-research (PyPI) < 1.6.10, ECOSYSTEM:introduced=0,fixed=1.6.10.
Is there a fix for CVE-2026-46526?
Yes. local-deep-research (PyPI): upgrade to 1.6.10 or later. Junglewise finds every place you run it, drafts the upgrade and applies it once you approve.
How is CVE-2026-46526 attacked?
It is attacked over the network, in a software library.
Am I affected by CVE-2026-46526?
Check whether you run local-deep-research (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.

How to fix

  • local-deep-research (PyPI): upgrade to 1.6.10 or later
    pip install "local-deep-research>=1.6.10"

Technical details

A Server-Side Request Forgery (SSRF) vulnerability exists in Local Deep Research due to a parser differential between the 'urllib.parse.urlparse' library used for validation and the 'urllib3' library used by 'requests' to fetch content. Specifically, URLs containing backslashes (e.g., http://127.0.0.1\@1.1.1.1) are parsed differently: the validator sees the external host (1.1.1.1), while the HTTP client connects to the internal host (127.0.0.1). This allows an authenticated attacker to bypass 'validate_url' checks and reach internal loopback services or cloud metadata endpoints (AWS, Azure, etc.). The vulnerability is fixed in version 1.6.10 by switching the validator to use the 'urllib3' parser and implementing stricter RFC 3986 character filtering.

Affected products

  • LearningCircuit local-deep-research < 1.6.10

Identifiers

  • CVE-2026-46526
  • GHSA-g23j-2vwm-5c25
  • PYSEC-2026-2611

Timeline

  • 2026-05-08: patched: Fixes merged into main branch via PR #3873
  • 2026-05-28: advisory: CVE-2026-46526 published

References

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