Threat database

CVE-2026-105148 is a server-side request forgery (SSRF) vulnerability in SciPhi-AI R2R versions 3.6.0 through 3.6.6. It affects the Retrieval Completion API Endpoint component, specifically the file py/shared/abstractions/llm.py. The vulnerability arises from manipulation of the argument generation_config.api_base, allowing remote attackers to induce the server to make unauthorized requests. Exploit code is publicly available, but no vendor response or patch has been provided.

CVE-2026-105147 is a medium severity vulnerability in SciPhi-AI R2R versions 3.6.0 through 3.6.6. It involves hard-coded credentials in the JWT Secret Handler component due to manipulation of default secret keys. The vulnerability can be exploited remotely without authentication. Although the exploit has been publicly disclosed, there are no known exploits in the wild and the vendor has not responded to the disclosure. No patch or remediation information is currently available.

R2R through 3.6.6 contains a SQL injection vulnerability that allows unauthenticated attackers to inject SQL predicates into the chunks search query by manipulating the filter key parameter in the retrieval search endpoint. Attackers can exploit the direct interpolation of filter keys into the SQL WHERE clause without parameterization or escaping to perform time-based and boolean-based data exfiltration from the application database.

R2R through 3.6.6 contains a stacked SQL injection vulnerability that allows unauthenticated attackers to execute arbitrary SQL statements by manipulating the index name parameter in the vector index creation endpoint. The index name is interpolated directly into a CREATE INDEX statement via string formatting without identifier quoting or allowlist validation, enabling arbitrary DDL and DML execution through semicolon-separated statements under the PostgreSQL superuser account.

R2R through 3.6.5 fails to properly validate user ownership in conversation update and message handlers, allowing authenticated users to modify other users' conversations. Attackers can supply arbitrary conversation identifiers to rename conversations and append messages to other users' conversation histories, corrupting state and injecting malicious content.