CVE-2026-44487: Axios Proxy Credential Leakage on Redirect – Node.js Vulnerability

Technical summary

The vulnerability exists in Axios's Node.js HTTP adapter during specific proxy-to-direct redirect transitions. When a client initializes an HTTP request with a Proxy-Authorization header (authenticating to an outbound proxy), and the server responds with a redirect (3xx status), Axios follows the redirect to the new URL. If that redirected URL is no longer destined for the proxy—meaning it should be accessed directly—the adapter may still forward the Proxy-Authorization header to the final origin. This violates the principle that proxy credentials should never be shared with non-proxy endpoints. The issue is classified under CWE-201 (Exposure of Sensitive Information to an Unauthorized Actor), reflecting the unintended disclosure of authentication material.

Business impact

Organizations using Axios in Node.js backend services, microservices, or data-collection applications that route traffic through authenticated proxies face credential compromise. If proxy credentials leak to an attacker-controlled or attacker-observable endpoint, the attacker may be able to impersonate legitimate traffic and access internal resources behind the proxy. This is particularly concerning in environments where proxy credentials grant access to restricted APIs, internal services, or gated infrastructure. Data exfiltration, unauthorized API calls, and lateral movement within corporate networks become possible if the proxy is a gateway to sensitive systems.

Affected systems

Node.js applications using Axios versions prior to 0.32.0 (for the 0.x branch) or 1.16.0 (for the 1.x branch) are affected. Browser-based Axios usage is not impacted because browsers do not natively support authenticated HTTP proxies in the same way. The risk is highest in backend services, serverless functions, CI/CD pipelines, and other server-side contexts where Axios is used to make outbound HTTP requests and redirect-following is enabled. Applications that do not route traffic through authenticated proxies, or that disable automatic redirect following, face minimal risk from this specific flaw.

Exploitability

Exploitation does not require authentication or special user interaction. An attacker needs only to control or observe a destination that an Axios client is redirected to. The attack flow is straightforward: trigger a redirect from a proxied resource to an attacker-controlled endpoint, and the proxy credentials will be forwarded automatically. No network access restrictions or complex prerequisites are required. However, the attacker must first cause or intercept the initial redirect—typically by controlling a server the victim's application connects to, or by performing a man-in-the-middle attack in cases where TLS is not properly validated. Once that condition is met, credential leakage is highly likely.

Remediation

Upgrade Axios to version 0.32.0 or later (for 0.x users) or 1.16.0 or later (for 1.x users). These releases include fixes to the HTTP adapter that prevent proxy credentials from being forwarded to non-proxy destinations. After patching, validate that your application's redirect behavior remains unchanged and that any custom proxy configurations are still functional. If immediate patching is not feasible, consider disabling automatic redirect following in Axios by setting `maxRedirects: 0` in request configuration, and manually handling redirects only to trusted, known endpoints. Be aware that this workaround may require code refactoring.

Patch guidance

1. Identify all Node.js projects and dependencies using Axios. Check package.json and dependency trees for versions prior to 0.32.0 and 1.16.0. 2. Update Axios: Run `npm update axios` or `yarn upgrade axios` to pull the patched version. Verify the installed version matches or exceeds the fixed versions. 3. Test in a staging environment: Confirm that applications using proxy-based requests and redirects continue to function correctly after the upgrade. 4. Deploy to production in a phased manner, monitoring for any unexpected changes in proxy behavior or request handling. 5. For monorepos or pinned dependencies, manually update version constraints and re-lock dependency files before deployment.

Detection guidance

Monitor for Axios usage in your codebase and inventory the versions in use. Security scanning tools and dependency analyzers (npm audit, Snyk, Dependabot) can flag vulnerable versions automatically. At runtime, look for suspicious outbound connections from Node.js services to unexpected endpoints, or signs that proxy credentials are being used outside of proxy contexts. Network monitoring can detect Proxy-Authorization headers in requests to non-proxy servers. Log analysis of proxy infrastructure may reveal credential leakage attempts if they are observed and logged. Code review should verify that redirect handling does not inadvertently expose sensitive headers.

Why prioritize this

This vulnerability is HIGH severity (CVSS 7.5) and deserves immediate attention because it enables unauthenticated credential theft with no user interaction required, and affects a ubiquitous library used across Node.js ecosystems. The attack surface is broad: any backend service using Axios with proxy authentication is at risk. The fix is straightforward and patch versions are readily available. Organizations should prioritize patching all affected instances within 1–2 weeks, especially those handling sensitive APIs or operating in zero-trust network architectures where proxy credentials are a critical control.

Risk score, explained

The CVSS 3.1 score of 7.5 (HIGH) reflects high impact (confidentiality of credentials) with low attack complexity and no user interaction required. The vector CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:N/A:N indicates network-based exploitation, low barrier to entry, no privilege escalations needed, and compromise of sensitive information (credentials). Integrity and availability are not directly impacted, which is why the score does not reach CRITICAL; however, disclosed credentials can enable subsequent unauthorized actions.

Frequently asked questions