Cybersecurity

Active scanning detected targeting LiteLLM and Starlette vulnerabilities

Key Takeaway

Threat actors are actively scanning for unauthenticated remote code execution vulnerabilities in LiteLLM and Starlette via specific API endpoints.

Summary

Security researchers at the SANS Internet Storm Center have observed increased scanning activity targeting LiteLLM and Starlette installations. The observed requests specifically target the '/mcp-rest/test/tools/list' endpoint, indicating an attempt to identify vulnerable services.

These scans aim to exploit two specific vulnerabilities, CVE-2026-42271 and CVE-2026-48710. According to research from Horizon3.ai, these flaws can be chained together to allow an unauthenticated attacker to achieve remote code execution (RCE) on the target system.

Vulnerability details and attack vector

  • Vulnerability Identifiers: The attack targets CVE-2026-42271 and CVE-2026-48710.
  • Attack Vector: Remote, unauthenticated exploitation via specific API paths.
  • Exploitation Method: Attackers can chain these two vulnerabilities to bypass security controls and execute arbitrary code.
  • Target Endpoint: Scans have been identified hitting the '/mcp-rest/test/tools/list' URI.

Why it matters

Developers and system administrators deploying LiteLLM or Starlette-based services must prioritize patching to prevent unauthorized remote access. Because these vulnerabilities allow for unauthenticated RCE, they pose a critical risk to AI-integrated infrastructure and web services.

What to do

  • Update LiteLLM and Starlette installations to the fixed versions as soon as they become available from the vendors.
  • Monitor logs for requests to the '/mcp-rest/test/tools/list' endpoint, which indicates active scanning for these vulnerabilities.
  • Restrict network access to LiteLLM and Starlette API endpoints to authorized users only until patches can be applied.

Original reference: isc.sans.edu →