Jul 9, 2026·6 min read·16 visits
The Micronaut HTTP client fails to strip sensitive headers like Authorization and Cookie during cross-origin redirects, allowing attackers to hijack sessions via open redirects or malicious endpoints.
An information disclosure vulnerability exists in the Micronaut Framework's HTTP client components. The client fails to clear sensitive authorization headers and cookies when following redirects across different origins. If an application using the vulnerable client communicates with an endpoint that issues a redirect to an external host, the client will forward the original credentials, leading to potential token theft and session hijacking.
The vulnerability affects the core HTTP client libraries of the Micronaut Framework, specifically io.micronaut:micronaut-http-client and io.micronaut:micronaut-http-client-core. These libraries provide Netty-based HTTP client implementations used widely in Micronaut microservices to execute outbound HTTP calls.
When configuring an application to use the default HTTP client, the framework automatically handles HTTP redirect status codes (such as 301, 302, 307, and 308). The client is designed to preserve request state and copy headers from the initial request onto the subsequent redirected request.
However, the client did not differentiate between same-origin and cross-origin redirect destinations. This behavior allowed sensitive session credentials, including HTTP Authorization tokens, Proxy-Authorization headers, and local session cookies, to be forwarded to untrusted, external third-party servers.
The root cause of this vulnerability lies in the implementation of the setRedirectHeaders method in the Netty-based HTTP client components (DefaultHttpClient and NettyHttpClient). Historically, the client utilized a static, hardcoded blocklist named REDIRECT_HEADER_BLOCKLIST to filter headers before dispatching a redirected request.
This static blocklist was defined as follows:
private static final HttpHeaders REDIRECT_HEADER_BLOCKLIST;
static {
REDIRECT_HEADER_BLOCKLIST = new DefaultHttpHeaders();
REDIRECT_HEADER_BLOCKLIST.add(HttpHeaderNames.HOST, "");
REDIRECT_HEADER_BLOCKLIST.add(HttpHeaderNames.CONTENT_TYPE, "");
REDIRECT_HEADER_BLOCKLIST.add(HttpHeaderNames.CONTENT_LENGTH, "");
REDIRECT_HEADER_BLOCKLIST.add(HttpHeaderNames.TRANSFER_ENCODING, "");
REDIRECT_HEADER_BLOCKLIST.add(HttpHeaderNames.CONNECTION, "");
}The blocklist only targeted transport-level headers and body-related properties. When preparing the redirect request, the code copied all other original headers without verifying if the target destination host matched the origin of the initial request.
Because the client lacked origin-awareness during redirect processing, any application transmitting bearer tokens or session cookies to a trusted service would unknowingly send those exact same headers to an external server if the trusted service responded with a cross-origin redirect instruction.
To address this flaw, the Micronaut maintainers refactored the HTTP client configuration to categorize and separate filtered headers based on the context of the redirect.
Specifically, the fix introduced class fields to track three separate groups of headers:
redirectAlwaysFilteredHeaders: Headers that are stripped regardless of redirect context (e.g., Host, Connection).redirectAdditionalNonPreserveBodyFilteredHeaders: Headers stripped specifically when the HTTP redirect method does not preserve the request body (e.g., Content-Length, Content-Type).redirectCrossOriginFilteredHeaders: High-sensitivity headers that must be stripped exclusively on cross-origin redirects (e.g., Authorization, Proxy-Authorization, Cookie).The following code comparison details the implementation changes:
// Vulnerable Implementation
private void setRedirectHeaders(@Nullable io.micronaut.http.HttpRequest<?> request,
MutableHttpRequest<Object> redirectRequest) {
if (request != null) {
for (Map.Entry<String, List<String>> originalHeader : request.getHeaders()) {
if (!REDIRECT_HEADER_BLOCKLIST.contains(originalHeader.getKey())) {
final List<String> originalHeaderValue = originalHeader.getValue();
if (originalHeaderValue != null && !originalHeaderValue.isEmpty()) {
for (String value : originalHeaderValue) {
if (value != null) {
redirectRequest.header(originalHeader.getKey(), value);
}
}
}
}
}
}
}// Patched Implementation
private void setRedirectHeaders(@Nullable io.micronaut.http.HttpRequest<?> request,
MutableHttpRequest<Object> redirectRequest,
boolean preserveBody) {
if (request == null) {
return;
}
boolean sameOrigin;
try {
// Determine if the redirect destination is the same origin
sameOrigin = new RequestKey(this, request.getUri())
.equals(new RequestKey(this, redirectRequest.getUri()));
} catch (Exception e) {
sameOrigin = false;
}
// Resolve blocklist based on destination origin alignment
DefaultHttpHeaders headersToBlock = resolveRedirectFilteredHeaders(!sameOrigin, preserveBody);
for (Map.Entry<String, List<String>> originalHeader : request.getHeaders()) {
String headerName = originalHeader.getKey();
if (headersToBlock.contains(headerName)) {
continue;
}
// Copy non-filtered headers securely...
}
}The logic relies on RequestKey equality to perform the origin check. If the origins do not match, !sameOrigin evaluates to true, forcing the retrieval and addition of the redirectCrossOriginFilteredHeaders to the active blocklist, which effectively prevents the leakage of sensitive credential headers.
An attacker can exploit this vulnerability through several vectors. The most direct approach is leveraging an open redirect vulnerability on a trusted target API, or compromising a trusted host to redirect traffic to an attacker-controlled endpoint.
The Micronaut client initiates an authenticated HTTP request containing an OAuth bearer token in the Authorization header to https://trusted-service.com/api.
The server at trusted-service.com yields a 302 Found response containing a Location header that points to https://attacker-controlled-site.com/log.
The vulnerable HTTP client receives the 302 response. Because it does not recognize https://attacker-controlled-site.com as cross-origin, it creates a new request and retains the original Authorization header containing the bearer token.
The client dispatches the request to the attacker's server, which logs all incoming HTTP request headers and captures the active token.
The impact of this vulnerability is rated as High severity. The leakage of sensitive transport credentials allows remote attackers to perform full session hijacking, account takeover, or unauthorized downstream operations.
Because many corporate enterprise architectures utilize centralized identity providers, OAuth 2.0 bearer tokens, or JWTs, a single compromised token can grant access to a wide array of connected backend services. If the vulnerable client is configured to communicate with internal microservices, the leaked credentials can enable lateral movement across internal network boundaries.
Furthermore, if the application runs within a cloud infrastructure and queries metadata services using custom tokens, a cross-origin redirect triggered by external input can lead to cloud-level privilege escalation.
While the official patches systematically block default authentication headers (Authorization, Proxy-Authorization, and Cookie), security teams must evaluate and manage residual risks related to custom authorization schemes.
Many API architectures use non-standard, custom HTTP headers to pass API keys or JWT tokens (e.g., X-Api-Key, X-Auth-Token, or X-Access-Token). The default patched implementation does not include these custom headers in the default redirectCrossOriginFilteredHeaders blocklist.
Consequently, applications using custom security headers remain vulnerable to credential leakage during cross-origin redirects unless developers explicitly configure the client to filter those specific custom header keys.
The definitive remediation is to upgrade the Micronaut framework dependencies to the patched versions.
64e539736b8168f201d868b02ace50fe14f57418.70cab4b44fbf985faba2846091f2356b5bd70719 and 9770328999f490bdfbb9e25addd45bf73d4a173a.If immediate dependency upgrades are not feasible, apply the following configurations to mitigate the risk:
micronaut:
http:
client:
follow-redirects: falsemicronaut:
http:
client:
redirect-cross-origin-filtered-headers:
- "Authorization"
- "Proxy-Authorization"
- "Cookie"
- "X-Api-Key"
- "X-Auth-Token"CVSS:3.1/AV:N/AC:L/PR:N/UI:R/S:U/C:H/I:H/A:N| Product | Affected Versions | Fixed Version |
|---|---|---|
micronaut-http-client Micronaut | < 3.10.4 | 3.10.4 |
micronaut-http-client-core Micronaut | < 3.10.4 | 3.10.4 |
micronaut-http-client Micronaut | >= 4.0.0, < 4.0.1 | 4.0.1 |
| Attribute | Detail |
|---|---|
| CWE ID | CWE-200 / CWE-522 |
| Attack Vector | Network (Cross-Origin Redirects) |
| CVSS Severity Score | 8.8 |
| EPSS Score | N/A |
| Exploit Status | PoC available in test suites |
| CISA KEV Status | Not Listed |
The product exposes sensitive information to an actor that is not explicitly authorized to have access to that information.
CVE-2026-12243 is a path traversal vulnerability in the Natural Language Toolkit (NLTK) version 3.9.4. The flaw exists because the input validation routine fails to account for percent-encoded directory traversal sequences like '..%2f' before passing them to urllib.request.url2pathname(), which decodes them into active traversal sequences.
CVE-2026-73654 is a high-severity prototype pollution vulnerability in Trigger.dev. The flaw occurs during the handling of run-metadata updates through the PUT /api/v1/runs/:runId/metadata endpoint. Because user-supplied keys are parsed directly by the @jsonhero/path library without sanitization, an authenticated attacker with low privileges can pollute the global Object.prototype. This causes database queries via Prisma ORM to fail validation and induces unhandled exceptions in the Prometheus metrics client, resulting in a process-wide denial of service.
CVE-2026-73559 is an uncontrolled resource consumption vulnerability in the vLLM engine, specifically within the /v1/completions API endpoint, allowing authenticated attackers to cause application-level denial of service via unbounded prompt arrays.
A critical security bypass vulnerability in Argo Workflows allows authenticated attackers with workflow submission privileges to bypass 'Strict' or 'Secure' template referencing restrictions. By injecting unvalidated fields into the nested ArtifactGC configuration, attackers can execute arbitrary pod patches, leading to host namespace escape and cluster-wide privilege escalation.
A Server-Side Request Forgery (SSRF) / Confused Deputy vulnerability has been identified in Pydantic AI UI Adapters (such as VercelAIAdapter). Under certain conditions, a malicious client can supply manipulated message history with provider metadata that forces the server to resolve files within privileged cloud environments (AWS S3, Google Cloud Storage) or model providers. This occurs because the adapters deserialize client-provided metadata structures directly into UploadedFile instances without validation, which are subsequently fetched using high-privilege server credentials.
A path traversal vulnerability in the optional dashboard server of atomic-agents-stack before version 1.1.0 allows unauthenticated remote attackers to read arbitrary files from the host filesystem.