CVE-2026-57143: CWE-20: Improper Input Validation in MervinPraison praisonaiagents

PraisonAI is a multi-agent teams system. In praisonaiagents versions prior to 1.6.62, the searxng_search and search_web SearxNG provider components do not validate the searxng_url parameter before passing it to requests.get. This lack of validation on scheme, host, port, and resolved address allows an attacker controlling the LLM input to direct the search tool to internal JSON services, distinguish open and closed internal ports, and attempt cloud metadata service requests. The returned JSON data is then incorporated into the agent's context, potentially leaking sensitive internal information. The vulnerability is classified under CWE-20 (Improper Input Validation) and CWE-918 (Server-Side Request Forgery). The issue is resolved in praisonaiagents version 1.6.62.

An attacker can exploit this vulnerability to perform server-side request forgery (SSRF) attacks by injecting crafted input that causes the system to make unauthorized requests to internal services, including cloud metadata endpoints. This can lead to information disclosure of internal network details and sensitive JSON data. The CVSS v3.1 score of 8.8 (high severity) reflects the potential for high confidentiality impact, limited integrity impact, and high availability impact. There are no known exploits in the wild at this time.

Upgrade praisonaiagents to version 1.6.62 or later, where the vulnerability is fixed by proper validation of the searxng_url parameter before use. No other mitigation is indicated or required according to the available data.