CVE-2026-54037: LibreChat Rate-Limit Bypass on Conversation Duplication Endpoint

Technical summary

CVE-2026-54037 is a rate-limiting bypass in LibreChat versions prior to 0.8.4-rc1. The POST /api/convos/fork endpoint was patched with forkIpLimiter and forkUserLimiter rate limiters in response to CVE-2025-7105, which addressed rapid-fire conversation duplication attacks. The POST /api/convos/duplicate endpoint, located in the same file and executing identical expensive database operations, was not similarly protected. An authenticated attacker can invoke the duplicate endpoint repeatedly to exhaust server resources, effectively negating the CVE-2025-7105 fix. The vulnerability is classified under CWE-770 (Allocation of Resources Without Limits or Throttling).

Business impact

This vulnerability enables authenticated users to conduct denial-of-service attacks against LibreChat deployments without triggering rate limits. In production environments, repeated resource exhaustion via the duplicate endpoint could degrade service availability, slow legitimate conversation processing, or overwhelm database connections. Organizations relying on LibreChat for internal AI interactions face potential uptime and performance risks. The impact is partially constrained by the requirement for user authentication, but the ease of exploitation and the completeness of the bypass—affecting the exact same operations—elevates business risk.

Affected systems

All LibreChat versions prior to 0.8.4-rc1 are vulnerable. The flaw affects any deployment that exposes the POST /api/convos/duplicate endpoint to authenticated users. Organizations running LibreChat should verify their current version and plan an upgrade to 0.8.4-rc1 or later.

Exploitability

Exploitability is straightforward. The vulnerability requires valid user authentication (authenticated access) but no special privileges or complex setup. An attacker with a user account can repeatedly call POST /api/convos/duplicate with varied conversation parameters to exhaust CPU, memory, and database resources. No user interaction or social engineering is needed. The attack is practical and scalable, making it a credible risk in multi-tenant or public LibreChat instances.

Remediation

Upgrade LibreChat to version 0.8.4-rc1 or later, where the duplicate endpoint receives identical rate-limiting protections as the fork endpoint. If immediate patching is not possible, consider implementing reverse-proxy or WAF-level rate limits on the /api/convos/duplicate endpoint to restrict authenticated requests by IP or user ID. Verify that both fork and duplicate endpoints are rate-limited symmetrically after deployment.

Patch guidance

LibreChat maintainers have resolved this issue in 0.8.4-rc1. Deploy this release or any subsequent stable version. Prior to applying the patch, review your LibreChat deployment logs for suspicious patterns of repeated calls to POST /api/convos/duplicate, which would indicate prior exploitation attempts. Test the patched version in a staging environment to confirm rate-limiting behavior on both endpoints before production rollout.

Detection guidance

Monitor application logs for repeated POST requests to /api/convos/duplicate from the same authenticated user or IP address within short time windows (e.g., >10 requests/minute). Alert on unusual spikes in database query duration or CPU consumption following patterns of duplicate endpoint calls. Network-based detection can identify anomalous request frequency to this endpoint using WAF rules or SIEM correlation. Baseline normal conversation duplication activity for your deployment and flag deviations.

Why prioritize this

Although the CVSS score is 6.5 (Medium), the vulnerability warrants timely attention because it explicitly defeats a prior security fix, indicating incomplete remediation practices by the maintainers. The ease of exploitation by any authenticated user, combined with direct denial-of-service potential, justifies prioritization above baseline medium-severity guidance. Upgrade should be scheduled within 1–2 weeks for most deployments.

Risk score, explained

The CVSS 3.1 score of 6.5 reflects a network-accessible (AV:N), low-complexity (AC:L) attack requiring user authentication (PR:L) with no confidentiality or integrity impact but high availability impact (A:H). The score appropriately captures the DoS risk but does not account for the context that this is a bypass of a prior fix—a signal of incomplete patching practices—which may warrant internal risk elevation in your threat model.

Frequently asked questions