Mar 20, 2026·5 min read·25 visits
OpenClaw gateway versions up to 2026.3.11 fail to strip client-declared scopes during WebSocket initialization if the client provides a valid shared token. This grants administrative access (e.g., operator.admin) to deviceless connections without server-side validation.
A logic flaw in the OpenClaw gateway WebSocket connection handler permits clients authenticating with shared tokens to self-declare and retain elevated administrative scopes. This vulnerability allows an attacker possessing a low-privileged shared secret to bypass intended device-identity boundaries and execute administrative RPC commands against the gateway.
The OpenClaw gateway relies on WebSocket connections to facilitate communication between client devices, the control UI, and backend services. During connection initialization, the gateway enforces an authorization model that binds specific access scopes to authenticated device identities or trusted control interfaces.
Versions of the openclaw npm package prior to 2026.3.12 contain a critical authorization bypass vulnerability within this WebSocket connection path. The issue arises from a logical error in how the gateway processes self-declared scopes provided by the client during the initial handshake.
The vulnerability is classified as Improper Privilege Management (CWE-269) and Missing Authorization (CWE-862). An attacker possessing a valid shared authentication token can exploit this flaw to assume arbitrary authorization scopes, completely bypassing the intended Role-Based Access Control (RBAC) mechanisms.
The root cause of the vulnerability resides in the attachGatewayWsMessageHandler function located in src/gateway/server/ws-connection/message-handler.ts. This function handles incoming WebSocket messages and manages connection state, including the assignment and validation of authorization scopes.
When a client connects, it can submit a requested array of scopes. The gateway is designed to strip these scopes using a clearUnboundScopes subroutine unless the connection is explicitly bound to a trusted Control UI operator or a strictly verified device identity.
The logic flaw occurs within the conditional statement that dictates when clearUnboundScopes executes. The vulnerable implementation skips scope clearing if the sharedAuthOk flag evaluates to true. This flag indicates only that the client successfully authenticated using a shared token or password, not that the client possesses the authorization required for the requested scopes.
By treating shared secret authentication as functionally equivalent to a trusted authorization binding, the gateway inherently trusts the client's self-declared scopes. This architectural confusion between authentication (verifying the shared token) and authorization (verifying the right to hold the operator.admin scope) directly causes the privilege escalation.
The vulnerable code in message-handler.ts defines the clearUnboundScopes function with a flawed conditional guard. If the client provides a valid shared token, sharedAuthOk is true, causing the function to exit without clearing the unbound scopes.
// Vulnerable Logic in src/gateway/server/ws-connection/message-handler.ts
const clearUnboundScopes = () => {
// BUG: If sharedAuthOk is true, scopes are NOT cleared.
if (scopes.length > 0 && !controlUiAuthPolicy.allowBypass && !sharedAuthOk) {
scopes = [];
connectParams.scopes = scopes;
}
};The patch implemented in commit 5e389d5e7c9233ec91026ab2fea299ebaf3249f6 addresses this by strictly separating the scope-clearing action from the authentication state. The clearUnboundScopes function was simplified to unconditionally clear scopes whenever invoked.
// Patched Logic
const clearUnboundScopes = () => {
if (scopes.length > 0) {
scopes = [];
connectParams.scopes = scopes;
}
};Furthermore, the patch relocates the authorization decision logic into the handleMissingDeviceIdentity block. Scopes are now stripped by default unless the connection definitively originates from an authorized Control UI path with explicit server-side trust.
// Authorization decision logic (Patched)
if (!device && (!isControlUi || decision.kind !== "allow")) {
clearUnboundScopes();
}Exploitation requires the attacker to possess a valid shared token or password for the OpenClaw gateway. This token provides basic authentication but is intentionally designed to lack administrative privileges when used without a paired device identity.
The attacker initiates a standard WebSocket connection to the gateway's endpoint. During the connection request, the attacker injects a JSON payload containing the shared token and an explicit declaration of elevated scopes, specifically operator.admin.
// Proof-of-Concept WebSocket Payload
const res = await connectReq(ws, {
token: "secret", // Known shared token
scopes: ["operator.admin"], // Self-declared elevated scope
device: null // Deviceless connection
});Upon processing this request, the vulnerable gateway verifies the token, sets sharedAuthOk = true, and skips the scope stripping subroutine. The server registers the connection state with the operator.admin scope applied.
The attacker can subsequently issue administrative Remote Procedure Calls (RPC) over the established WebSocket tunnel. For example, the attacker can submit a set-heartbeats RPC command to disable monitoring, which the server processes successfully due to the retained administrative scope.
The vulnerability carries a CVSS v3.1 score of 9.9, reflecting critical severity. The attack vector is network-based, requires low privileges (a standard shared token), and does not require user interaction.
Successful exploitation results in a complete bypass of the gateway's authorization model. An attacker achieves operator.admin access, granting them the ability to manage gateway configurations, modify heartbeat settings, and potentially pivot into internal administrative functions exposed via the WebSocket RPC interface.
The scope change (S:C) in the CVSS vector highlights that the vulnerability allows a low-privileged client component to manipulate the security posture of the central gateway infrastructure. This compromises the confidentiality, integrity, and availability of the system managed by OpenClaw.
The primary remediation for this vulnerability is to upgrade the openclaw package to version 2026.3.12 or later. The patch structurally resolves the logic flaw by enforcing strict server-side scope bindings and rejecting client-declared scopes for deviceless connections.
Organizations unable to patch immediately should rotate all shared authentication tokens and passwords. While this does not fix the underlying vulnerability, it prevents exploitation by attackers who may have obtained standard credentials through prior leaks or unauthorized access.
Network security teams should implement monitoring on the WebSocket traffic. Detection strategies should focus on identifying connection initialization payloads that attempt to claim operator.admin or similar sensitive scopes without an accompanying cryptographic device identity token.
CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:C/C:H/I:H/A:H| Product | Affected Versions | Fixed Version |
|---|---|---|
openclaw OpenClaw Ecosystem | <= 2026.3.11 | 2026.3.12 |
| Attribute | Detail |
|---|---|
| CWE ID | CWE-269 |
| Attack Vector | Network |
| CVSS Score | 9.9 (Critical) |
| Exploit Status | Proof of Concept |
| Authentication Required | Low (Shared Token) |
| User Interaction | None |
Improper Privilege Management
The @jhb.software/payload-cloudinary-plugin exposes an endpoint that performs unvalidated cryptographic signing of Cloudinary API parameters, allowing authenticated users with minimal privileges to forge valid signatures for arbitrary actions. This flaw allows attackers to overwrite remote storage assets, execute unauthorized file uploads, alter asset visibility parameters, trigger SSRF webhooks, and perform directory traversal within Cloudinary repositories.
A Server-Side Request Forgery (SSRF) and Bearer Token Exfiltration vulnerability exists in the @merill/lokka (Lokka) Model Context Protocol (MCP) server prior to version 2.1.2. The server constructed Azure Resource Manager request URLs by concatenating user-controlled path parameters directly into destination request strings. By injecting authority-redefinition characters, an attacker can manipulate URL parsing to execute a host-escape attack, forcing the server to send high-privilege Azure Resource Manager (ARM) Bearer tokens to an external attacker-controlled host. This allows complete administrative access to the associated Azure subscriptions.
A directory traversal and symlink following vulnerability exists in Pydantic Settings when using the NestedSecretsSettingsSource with nested subdirectory lookups enabled. An attacker capable of writing to the secrets directory can bypass size limitations, read arbitrary host files, or cause a denial-of-service condition via cyclic symlinks.
A Server-Side Request Forgery (SSRF) vulnerability exists in SurrealDB's Identity & Access Management (IAM) module prior to version 3.1.5. When configuring JSON Web Key Set (JWKS) URLs for token verification, the remote fetcher follows HTTP redirects by default without validating redirect targets against configured network capabilities. This allows high-privileged users to bypass network access limits and perform blind port scanning of internal network resources.
A local file disclosure vulnerability exists in SurrealDB's full-text search capabilities, allowing authenticated users with database EDITOR or OWNER roles to read arbitrary files from the host system filesystem. This occurs by abusing the mapper() filter inside a DEFINE ANALYZER statement to point to system files.
SurrealDB versions 3.0.0 through 3.1.4 contain an information exposure vulnerability (CWE-203) where the query planner optimizes sorted queries using indexes on fields with field-level SELECT restrictions. Because the query planner performs index-based sorting before enforcing permission-based redaction, unauthorized users can observe the physical order of returned rows to deduce the relative values of protected fields.