CVE-2026-62240: CrewAI SSRF filter bypass in scraping tools

CVE-2026-62240 · Severity: high · CVSS 7.4 · Published 2026-07-13

CrewAI's web scraping tools include a URL validator that checks if URLs are safe before fetching them, but only performs the validation once before following redirects. An attacker can supply a URL that initially appears safe but redirects to internal network addresses or cloud metadata endpoints, allowing unauthorized access to sensitive internal services and configuration data.

Technologies: crewai-tools (PyPI). Vendors: PyPI.

Quick answers

Is CVE-2026-62240 exploited in the wild?
No. Junglewise Threat Intelligence has no report of CVE-2026-62240 being exploited in the wild.
How severe is CVE-2026-62240?
High severity, with a CVSS score of 7.4, published 13 July 2026.
Which products and versions are affected?
crewAIInc crewAI < 1.15.1; crewai-tools (PyPI) < 1.15.1, ECOSYSTEM:introduced=0,fixed=1.15.1.
Is there a fix for CVE-2026-62240?
Yes. crewai-tools (PyPI): upgrade to 1.15.1 or later. Junglewise finds every place you run it, drafts the upgrade and applies it once you approve.
How is CVE-2026-62240 attacked?
It is attacked over the network, in a software library.
Am I affected by CVE-2026-62240?
Check whether you run crewAIInc crewAI or crewai-tools (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

  • crewai-tools (PyPI): upgrade to 1.15.1 or later
    pip install "crewai-tools>=1.15.1"

Technical details

The vulnerability is a Server-Side Request Forgery (SSRF) bypass in the validate_url function of CrewAI's scraping toolset. The vulnerable component performs DNS resolution and blocklist checking on the initial URL but does not re-validate redirect targets, allowing attackers to chain a public-facing URL that redirects to restricted internal addresses (127.0.0.1, 169.254.169.254 cloud metadata, etc.) or use DNS rebinding techniques where a domain resolves differently on subsequent lookups. The attack vector is network-based with no authentication required and only passive user interaction (following a redirect). The fix (version 1.15.1 and later) implements a safe_get() helper that validates both the initial URL and every redirect target before following redirects, and restricts credential forwarding across origin boundaries.

Affected products

  • CrewAI crewai-tools < 1.15.1

Identifiers

  • CVE-2026-62240
  • GHSA-mr4r-hcgx-8p4h
  • PYSEC-2026-3819

Timeline

  • 2026-07-14: disclosed
  • 2026-06-26: patched: Fix merged in PR #6331; release 1.15.1 published 2026-06-27

References

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