CVE-2026-57143: PraisonAI SearxNG URL validation bypass in search tools

CVE-2026-57143 · Severity: high · CVSS 8.8 · Published 2026-10-12

PraisonAI is a multi-agent AI system that uses web search tools to gather information for intelligent agents. Before version 1.6.62, an attacker can inject malicious search URLs through prompt injection that bypass security checks, allowing the system to make requests to internal services, scan for open ports, or attempt to access cloud metadata. This gives attackers visibility into the target's internal network infrastructure and the ability to extract sensitive information.

Technologies: MervinPraison Praisonaiagents, praisonaiagents (PyPI). Vendors: MervinPraison, PyPI.

Quick answers

Is CVE-2026-57143 exploited in the wild?
No. Junglewise Threat Intelligence has no report of CVE-2026-57143 being exploited in the wild.
How severe is CVE-2026-57143?
High severity, with a CVSS score of 8.8, published 12 October 2026.
Which products and versions are affected?
MervinPraison Praisonaiagents before 1.6.62; praisonaiagents (PyPI) ECOSYSTEM:introduced=0,fixed=1.6.61.
Is there a fix for CVE-2026-57143?
Yes. praisonaiagents (PyPI): upgrade to 1.6.61 or later. Junglewise finds every place you run it, drafts the upgrade and applies it once you approve.
How is CVE-2026-57143 attacked?
It is attacked over the network, in a software library.
Am I affected by CVE-2026-57143?
Check whether you run MervinPraison Praisonaiagents or praisonaiagents (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

  • praisonaiagents (PyPI): upgrade to 1.6.61 or later
    pip install "praisonaiagents>=1.6.61"

Technical details

The searxng_search and search_web SearxNG providers pass the LLM-controlled searxng_url parameter directly to requests.get without validating the URL scheme, host, port, or resolved address. An attacker can inject arbitrary URLs through prompt injection to scan internal services, enumerate port availability, or access cloud metadata endpoints; returned JSON results are incorporated into agent context. The fix is available in praisonaiagents 1.6.62.

Affected products

  • MervinPraison praisonaiagents before 1.6.62

Identifiers

  • CVE-2026-57143
  • GHSA-4pcv-mg8v-vrgf
  • PYSEC-2026-3530

Timeline

  • 2026-10-12: disclosed

References

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