HIGH 8.6

CVE-2026-55638: 9Router Authentication Bypass Vulnerability (CVSS 8.6)

9Router, an AI-based router designed to optimize and reduce token consumption for language model API calls, contains an authentication bypass vulnerability in versions before 0.5.2. The vulnerability exists because the application's URL protection rules in its dashboard guard don't cover the /codex endpoint. An attacker can exploit this gap by sending unauthenticated requests directly to /codex paths, which are later rewritten internally to /api/v1/responses. This rewriting happens after the authentication checks, allowing the attacker to bypass API-key verification and trick the server into making requests to upstream LLM providers using the operator's stored credentials. The impact ranges from data exposure to potential service disruption.

Source data · NVD / CISA · public domain

CVSS
3.1 · 8.6 HIGH · CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:L/I:L/A:H
Weaknesses (CWE)
CWE-862, CWE-863
Affected products
0 configuration(s)
Published / Modified
2026-07-10 / 2026-07-10

NVD description (verbatim)

9Router is an AI router & token saver. Prior to 0.5.2, 9router protects /v1, /v1beta, /api/v1, and /api/v1beta in src/dashboardGuard.js but omits /codex before next.config.mjs rewrites /codex/* to /api/v1/responses. A remote unauthenticated attacker can send requests to /codex/* to bypass the API-key gate and cause the server to make upstream provider calls using operator-stored LLM provider credentials. This issue is fixed in version 0.5.2.

4 reference(s) · View on NVD →

SEC.co analysis · AI-assisted, reviewed against source

Technical summary

The vulnerability arises from incomplete route coverage in src/dashboardGuard.js, which protects /v1, /v1beta, /api/v1, and /api/v1beta endpoints but excludes /codex. The next.config.mjs file contains a rewrite rule that transforms /codex/* requests to /api/v1/responses. Because this rewrite occurs after authentication checks, an unauthenticated attacker can reach protected functionality by sending requests to the unprotected /codex path. Once rewritten, these requests execute with the operator's LLM provider credentials (stored server-side), enabling unauthorized API calls to external providers. The CWE classifications (CWE-862: Missing Authorization and CWE-863: Incorrect Authorization) reflect the authentication bypass and authorization logic gap.

Business impact

A successful exploit allows attackers to make unauthorized calls to upstream LLM providers on behalf of the operator, potentially incurring substantial costs through token consumption. Depending on the provider's billing model and the attacker's volume, this could result in unexpected charges or service exhaustion. Additionally, the operator's stored credentials may be further compromised or exposed through verbose error messages or logging. Organizations running 9Router as part of their LLM infrastructure face both financial and reputational risk if an attacker generates suspicious API activity in their provider account.

Affected systems

The vulnerability affects 9Router versions prior to 0.5.2. No specific vendor or product count was disclosed in the advisory. Any organization deploying 9Router as a middleware layer for LLM API traffic is potentially at risk if running an unpatched version. The vulnerability has no bearing on upstream LLM providers (OpenAI, Anthropic, etc.) themselves—it is specific to 9Router's authentication layer.

Exploitability

Exploitability is straightforward and requires no authentication, special privileges, or user interaction. An attacker only needs to craft HTTP requests to /codex/* endpoints and can immediately bypass the API-key requirement. The vulnerability is accessible over the network with minimal complexity; there are no additional client-side requirements or environmental conditions. This makes the attack surface broad for any publicly exposed 9Router instance.

Remediation

Upgrade 9Router to version 0.5.2 or later. The fix applies authentication and authorization checks to the /codex endpoint before any URL rewrites occur, ensuring that all protected routes are covered consistently. Operators should verify the upgrade has been applied and test that unauthenticated /codex requests are now rejected. As an interim measure on older versions, network-level access controls or WAF rules can restrict traffic to /codex paths, though this is not a substitute for the patch.

Patch guidance

Update 9Router to version 0.5.2 or newer as soon as possible. Consult the official 9Router repository or release notes to confirm availability and deployment instructions for your environment. If you are running 9Router in a containerized or cloud-deployed setup, ensure the image or service definition pulls the patched version. Test the update in a staging environment first to confirm no downstream integrations are affected, though the patch is expected to be backward-compatible.

Detection guidance

Monitor your 9Router instance logs for repeated or unusual requests to /codex/* endpoints, especially from external or unexpected IP addresses. Check for a pattern of successful responses (HTTP 200, 201) followed by downstream API calls to LLM providers that don't correlate with your normal usage. If you have LLM provider billing alerts enabled, look for sudden spikes in API usage or token consumption immediately after your 9Router instance was publicly exposed. Review stored credentials access logs if available. A security audit of your LLM provider account's recent API activity is recommended even after patching, to identify and reverse any unauthorized calls.

Why prioritize this

This vulnerability merits immediate attention due to its high CVSS score (8.6), zero authentication requirement, and direct financial and operational impact. The simplicity of exploitation combined with the ease of chaining it to unauthorized API calls on behalf of the operator creates a critical risk. Organizations should treat this as urgent priority given the potential for rapid, large-scale abuse if 9Router is network-accessible.

Risk score, explained

The CVSS 3.1 score of 8.6 (HIGH) reflects a network-accessible vulnerability with low attack complexity, no authentication required, and high availability impact (through token exhaustion or service disruption). The integrity and confidentiality impacts are lower but present, as the attacker can trigger unauthorized actions and potentially extract information through error responses or logs. The score appropriately captures the balance between ease of exploitation and severity of outcome in this context.

Frequently asked questions

Can this vulnerability affect my LLM provider directly (e.g., OpenAI, Anthropic)?

No. This vulnerability is specific to 9Router's authentication layer. Your LLM provider's own security is not compromised. However, the attacker can abuse your provider account credentials stored in 9Router, potentially damaging your reputation and incurring costs on your behalf.

How can I tell if my 9Router instance has been exploited?

Check 9Router logs for requests to /codex/* paths from untrusted sources and correlate them with unusual LLM API calls in your provider's audit logs or billing alerts. Look for API usage that exceeds your expected patterns or calls at times when your applications were not active.

Do I need to rotate my LLM provider credentials after patching?

It is prudent to rotate credentials if your 9Router instance was public-facing before the patch was applied, as attackers may have already accessed or logged them. At minimum, review your LLM provider's recent API activity for unauthorized calls and reverse any charges if possible.

Will updating to version 0.5.2 break my existing integrations?

The patch is designed to be backward-compatible. However, always test in a staging environment first. If you have any integrations that were using /codex paths without API keys (relying on the bypass), they will now require proper authentication and will need to be updated to include valid API credentials.

This analysis is provided for informational purposes based on publicly available vulnerability data. SEC.co does not operate, endorse, or maintain 9Router. Verify all remediation steps against official 9Router documentation and advisories before implementation. Organizations should conduct their own risk assessment based on their specific deployment and threat model. This information is current as of the published date but may require updates as new details emerge. Source: NVD (public-domain), retrieved 2026-08-17. Analysis generated by SEC.co (claude-haiku-4-5).