External risk intelligence

R2R Stacked SQL Injection via Vector Index Creation Endpoint

CVE advisorySeverity: CRITICAL (CVSS 9.3)

CVE-2026-82526

The vulnerability resides in a vector index creation endpoint within an R2R application. As R2R is frequently deployed as an internet-facing API or web service for LLM applications, this functionality is commonly exposed to network-based interaction, making it a likely target for remote access in typical deployment patterns.

SQL Injection

Halo Surface Signal: 4 out of 5 — likely to be public-facing.

External exposure likelihood

Horizon Alert

Summary of the vulnerability and why it matters

This vulnerability allows unauthenticated attackers to execute arbitrary SQL commands by manipulating a specific parameter when creating indexes, potentially leading to unauthorized data access or modification. The concern is confirming relevance and exposure given the technical nature of the exploit.

Unauthenticated attackers can run database commands.
Matters if external access is used for index creation.
Confirm if our systems create indexes externally.

Attack Path

How an attacker could exploit the issue

An unauthenticated attacker can exploit a stacked SQL injection vulnerability in the R2R vector index creation endpoint by manipulating the index name parameter. This allows the attacker to execute arbitrary SQL statements, potentially leading to full control of the PostgreSQL superuser account and the underlying database.

Unauthenticated network access required.
Triggered by manipulating index name parameter.
Leads to arbitrary SQL execution.

Live Threat

Current exploitation, exposure, and threat context

This vulnerability could allow unauthenticated attackers to execute arbitrary SQL commands on the database. This is possible by manipulating the index name parameter when creating a vector index, which is directly inserted into a SQL statement without proper sanitization. When conditions are right, this could lead to the execution of SQL commands as the PostgreSQL superuser.

Database data and system commands at risk.
Via index name parameter manipulation.
Full system compromise possible.

Operational Fix

Recommended remediation, mitigation, and detection steps

The stacked SQL injection vulnerability in R2R affects applications that handle vector index creation, potentially exposing them to unauthenticated attackers. Application owners, in conjunction with platform and security teams, should prioritize identifying instances of R2R, confirming their exposure and criticality, and then planning remediation.

Application owners must identify R2R instances.
Verify network reachability and business criticality.
Plan remediation based on assessed risk.

Supplementary metadata

CVSS vector

Validate whether this threat affects your internet-facing exposure.

Halo Threat Intelligence helps prioritize remediation with Halo Surface Signal and H/A/L/O context. Start exposure validation with a free external attack surface trial.

Free EASM Trial See How Signal Works

Frequently asked questions

What is R2R and how is it used?

R2R is a software framework used to build Retrieval-Augmented Generation (RAG) systems. Developers use it to manage and search vector databases, which store high-dimensional data points. By indexing information, R2R enables LLM applications to retrieve relevant context quickly, effectively serving as the infrastructure that bridges raw data with AI models.

What does CVE-2026-82526 mean?

This CVE identifies a stacked SQL injection vulnerability (CWE-89). It occurs when the software takes user-provided input and embeds it directly into a database command without proper cleaning. In this specific case, the software fails to validate or quote the index name, which allows an attacker to inject and run their own, unauthorized SQL commands alongside the intended index creation task.

How does an attacker trigger this vulnerability?

An attacker exploits this by sending a specially crafted index name to the vector index creation endpoint. Because the application processes this input insecurely, it allows for 'stacked' queries, where the attacker ends their command with a semicolon to run additional, malicious database instructions. Simply navigating to the site or viewing data does not trigger this; the attacker must specifically interact with the index creation function.

Is my R2R instance at risk?

Halo Surface Signal indicates that R2R is often deployed as an internet-facing API, which frequently exposes its index creation functionality to remote network requests. If your R2R instance is accessible via the public internet or an untrusted network, it is at higher risk of being targeted. Internal-only instances that are not exposed to such traffic face a lower probability of unauthorized remote exploitation.

What should I do to address this issue?

First, locate all running instances of R2R within your environment. Once identified, evaluate whether these services allow external users to access the vector index creation endpoint. Work with your technical team to restrict network access to this endpoint immediately and prioritize updating or patching the affected R2R software to eliminate the insecure handling of the index name parameter.

References

Written by Nicholas Merritt

Nicholas Merritt is VP of Security Solutions at Halo Security. With more than 25 years across application security, vulnerability prioritization, and offensive security, he helps teams translate threat data into practical exposure validation and remediation focus. He authors Halo Threat Intelligence CVE advisories using Halo Surface Signal and H/A/L/O context to prioritize internet-facing risk.