The mem0 component exposes unauthenticated configuration API endpoints that leak large language model (LLM) API keys in plaintext. Specifically, the GET /api/v1/config/ endpoint allows retrieval of stored secrets without authentication. Additionally, the PUT /api/v1/config/mem0/llm endpoint accepts an attacker-controlled ollama_base_url parameter, which can be manipulated to perform server-side request forgery (SSRF) attacks targeting internal network resources such as cloud instance metadata services (IMDS). This vulnerability is tracked as CVE-2026-59706 and is classified under CWE-306 (Missing Authentication for Critical Function). No patch or official remediation has been provided in the available data.

Unauthenticated attackers can obtain sensitive API keys in plaintext, potentially compromising LLM service credentials. The SSRF capability allows attackers to make unauthorized requests from the vulnerable server to internal services, which may lead to further internal network compromise or data leakage. The combination of secret exposure and SSRF significantly increases the risk of unauthorized access and lateral movement within affected environments.

Patch status is not yet confirmed — check the vendor advisory for current remediation guidance. Until an official fix is available, restrict access to the affected API endpoints by implementing network-level controls or authentication proxies to prevent unauthenticated access. Monitor for unusual requests targeting the config API endpoints and consider disabling or restricting the ollama_base_url configuration if possible.