In this article The CaptiveCrunch campaign Storm-2945 and Midnight Blizzard CaptiveCrunch tradecraft and tooling How to protect against CaptiveCrunch activity Microsoft Defender detections and hunting guidance Indicators of compromise Since early May 2026, Microsoft Threat Intelligence has observed Storm-2945, a sub-cluster of Midnight Blizzard, conducting widespread but targeted traffic manipulation attacks involving hospitality sector networks served by captive portals worldwide. Despite some tactic, technique, and procedure (TTP) similarities to the Forest Blizzard DNS hijacking operation that we publicly disclosed in April 2026, we attribute this campaign, which we call CaptiveCrunch, to Storm-2945. As reported by ReliaQuest on July 23, a portion of this activity leverages doppelganger domains mimicking Microsoft online services to conduct follow-on adversary-in-the-middle (AitM) phishing operations that abuse the device code authentication flow in Microsoft Entra ID. Microsoft Threat Intelligence has also identified active traffic manipulation attacks leading to the delivery of malware on impacted systems. Microsoft has observed Storm-2945 leveraging AI to support a significant portion of these operations. Today, we are sharing our findings on these ongoing intrusions to raise awareness of this threat and enable customers to protect their devices, especially while traveling. We provide our assessment of Storm-2945’s relationship to Midnight Blizzard and analysis of the CaptiveCrunch campaign, detailing the malware and tradecraft used in these operations. We also provide mitigation, detection, and hunting guidance to help organizations identify and defend against Storm-2945 and related activity. Microsoft Threat Intelligence would like to thank our partners at Anthropic and OpenAI for their collaboration and support during this investigation. The CaptiveCrunch campaign Since February 2026, Storm-2945 has conducted AI-augmented operations including targeted device code and OAuth code phishing campaigns leading to Entra device registration and subsequent data collection from Microsoft 365. Since early May 2026, Microsoft Threat Intelligence has observed Storm-2945 manipulating DNS and HTTP traffic from networks served by captive portals to redirect user traffic through actor-controlled infrastructure. Although our investigation into the initial compromise vector for the captive portal networks is ongoing, we have observed notable commonalities in the equipment and management systems used across multiple affected networks. These similarities suggest that the activity might not be limited to isolated compromises of individual venues and could reflect access to shared services within portions of the captive portal ecosystem. Figure 1. Overview of the CaptiveCrunch attack flow As part of the CaptiveCrunch campaign, Storm-2945 has leveraged their AitM position to redirect users through actor-controlled phishing infrastructure and has also delivered malware purporting to be browser or operating system updates in response to automated connectivity checks issued by users’ browsers. Multiple variants have been delivered, including fully-featured Windows remote access trojans (RAT) in compiled Golang, with functionality to conduct system enumeration, collect files and keystrokes, steal credentials and session tokens, conduct audio and video surveillance, monitor for removable media, and provide the threat actor a remote shell on infected systems. The threat actor infrastructure leverages a variety of ClickFix techniques to elicit the user into downloading and executing the malware: Figure 2. ClickFix prompt with manual user instructions Figure 3. ClickFix prompt with additional user instructions after verification failure In addition to variants of malware targeting Windows systems, Microsoft Threat Intelligence is also aware of indications that the threat actor might be targeting Android devices with similar techniques as the ClickFi…
In SERVER mode, pgAdmin 4 enforces authentication per route via the @pga_login_required decorator; the application's before_request hook only handles desktop-mode auto-login and the Kerberos/Webserver-auth redirect, so any route shipped without the decorator is reachable without authentication (CWE-306). This is the same defect class previously fixed as CVE-2026-12046 (the sqleditor close/update_connection routes).
A follow-up sweep, prompted by a report describing an incomplete fix for CVE-2026-12046, found further routes missing @pga_login_required: the Constraints blueprint's nodes and proplist (object listing) routes and its delete route (a state-mutating DELETE that removes table constraints); preferences.get_all_cli (GET, discloses all CLI-settable preference values); debugger.close (DELETE); and schema_diff.close (DELETE). An unauthenticated network client could therefore enumerate constraint metadata, delete table constraints, read preference values, and force-close debugger or schema-diff sessions belonging to other users, without ever authenticating.
Fix adds the missing @pga_login_required decorator (and the corresponding import to the Constraints module) to each of these routes. The change is decorator-only; no behavioral changes to the underlying handlers.
This issue affects pgAdmin 4 in SERVER mode: the Constraints and Debugger routes from 1.0, the Schema Diff close route from 4.18, and preferences.get_all_cli from 8.2, all before 9.17.
Serendipity before 2.6.1 contains an open redirect vulnerability in exit.php that allows unauthenticated attackers to redirect users to arbitrary external sites by supplying a malicious Base64-encoded url parameter when the Track Exits plugin is configured with commentredirection set to s9y. Attackers can craft trusted-looking URLs leveraging the legitimate blog domain to conduct phishing, deliver malware, or bypass URL reputation filters.
A flaw was found in the keycloak-services component of Keycloak, which handles OpenID Connect (OIDC) authentication flows. The issue occurs because the security check designed to prevent HTTP parameter pollution only inspects the query portion of a redirect URL and ignores the fragment portion. When a client is configured with a wildcard redirect URI, an attacker can use this to inject duplicate security parameters into the login response. If a client application is not configured correctly, it might trust the attacker's injected data instead of the real security information from Keycloak, leading to session fixation or account confusion.
A flaw was found in the secure-client-uris client policy executor within Keycloak core services. This component is responsible for enforcing security requirements on client configurations, such as requiring encrypted connections for redirect URIs. Due to an improper check that only looks at the start of a web address rather than properly verifying the host, an attacker can bypass these security restrictions by using a specially crafted domain name. This could allow an attacker to intercept sensitive authentication codes over unencrypted connections.
A Server-Side Request Forgery (SSRF) vulnerability exists in Google mcp-toolbox versions 0.3.0 through 1.4.0. The vulnerability arises because the HTTP client used by the toolbox does not enforce restrictive redirect policies or validate target IPs, allowing crafted requests to cause unauthorized internal or external requests. This can lead to information disclosure or unauthorized interactions with internal systems. The vulnerability has a CVSS score of 6.1, indicating a high severity level.
MaxKey through 4.1.12, fixed in commit ddbb72f, contains an insufficient redirect URI validation vulnerability in DefaultRedirectResolver.hostMatches() that allows remote attackers to hijack OAuth 2.0 authorization codes by supplying a crafted redirect_uri whose hostname suffix matches a registered URI without proper dot-boundary anchoring. Attackers who control a domain ending with the registered redirect URI hostname can social-engineer victims into clicking a crafted authorization URL, causing the authorization code to be issued to the attacker-controlled URI and exchanged for an access token granting access to the victim's identity.
CVE-2026-53669 is an open redirect vulnerability in React Router, a routing library for React applications. A remote attacker can exploit this flaw by crafting malicious links using backslashes in navigation components, causing users to be redirected to arbitrary external websites. This vulnerability affects multiple Red Hat products that include React Router components. The issue is rated medium severity and could facilitate phishing attacks. Red Hat has released updated RPM packages for prometheus3.5 as part of their Hardened Images security update to address this vulnerability. No mitigation other than applying the update is currently recommended.
VaahCMS versions 2.0.0 through 2.3.4 contain a malicious obfuscated JavaScript payload embedded in the Blade template responsible for rendering security OTP emails, allowing remote attackers to execute unauthorized code in any browser that renders the affected email template with JavaScript enabled. The payload establishes a WebSocket connection to a hardcoded command-and-control endpoint, installs a password-field keylogger using MutationObserver to capture dynamically added inputs, scrapes WhatsApp Web DOM content, and accepts remote commands to redirect or overwrite the rendered page.
CVE-2026-67436 is a high-severity vulnerability in Linuxfabrik monitoring-plugins version 6.0.0 and earlier. The redfish-* plugins construct request URLs by concatenating an operator-supplied base URL with response-supplied @odata.id links. This improper input validation allows a malicious or compromised Baseboard Management Controller (BMC) to redirect authenticated Redfish requests, potentially exposing X-Auth-Token or HTTP Basic credentials.
Dify AI Workflow oauth_redirect_url Open Redirect Vulnerability. This vulnerability allows remote attackers to disclose sensitive information on affected installations of Dify. User interaction is required to exploit this vulnerability in that the target must visit a malicious page or open a malicious file.
The specific flaw exists within the OAuth flow handling in the AppInitializer component. An attacker can force a redirection to a site that serves malicious content. An attacker can leverage this vulnerability to disclose information in the context of the application. Was ZDI-CAN-29196.
Apache Traffic Server allows redirect-limit bypass when plugins reset the retry counter, enabling SSRF amplification.
This issue affects Apache Traffic Server: from 8.0.0 through 8.1.11, from 9.0.0 through 9.2.14, from 10.0.0 through 10.1.3.
Users are recommended to upgrade to version 9.2.15 or 10.1.4, which fix the issue.
Apache Traffic Server copies the client Host header into a fixed-size stack buffer without a bound during redirect handling, so an over-long Host header overflows the stack when redirect following is enabled.
This issue affects Apache Traffic Server: from 8.0.0 through 8.1.11, from 9.0.0 through 9.2.14, from 10.0.0 through 10.1.3.
Users are recommended to upgrade to version 9.2.15 or 10.1.4, which fix the issue.
Appium Java Client is the Java language binding for writing Appium tests that conform to the W3C WebDriver protocol. From 8.2.1 until 10.1.1, when directConnect(true) is enabled, AppiumCommandExecutor.setDirectConnect() reads the directConnectHost, directConnectPort, and directConnectPath fields from the server's NEW_SESSION response and rebuilds the client's server URL from them, validating only that the protocol is https, with no host allowlist or IP validation; a rogue or compromised server can therefore redirect all subsequent session traffic to an arbitrary destination, enabling full interception of session traffic and a server-side request forgery pivot to internal hosts, including cloud metadata (IMDS) credential theft. This vulnerability is fixed in 10.1.1.
A Server-Side Request Forgery (SSRF) vulnerability exists in Next.js versions 14.1.1 through 15.5.20 and 16.0.0 through 16.2.10. The flaw allows an attacker to cause the server to send outbound requests to malicious hosts by controlling Host-associated headers when Server Actions forward or redirect requests. This affects applications where the incoming host header is not fixed to a trusted value, typically in custom server setups or deployments not behind a host-pinning proxy. Managed hosting and certain Next.js configurations from version 14.2 onward are not affected. The issue has been fixed in versions 15.5.21 and 16.2.11.
Papra is a minimalistic document management and archiving platform. Prior to version 26.5.0, Papra's webhook delivery system contains an SSRF protection bypass that allows any authenticated organisation member to cause the server to make HTTP requests to internal addresses — loopback, link-local, and RFC-1918 ranges. The SSRF protection validates the registered webhook URL but ignores redirect destinations. The HTTP client (ofetch) follows 3xx responses automatically, and the redirect target is never checked against the blocklist. An attacker registers a webhook pointing to an attacker-controlled server, which redirects incoming POSTs to any internal address. Exploitation was confirmed by live test against the official Docker image. The fix is a single-line change to the webhook HTTP client. This issue has been patched in version 26.5.0.
This security update for agama-web-ui addresses multiple vulnerabilities across various dependencies, including unsafe use of Math.random(), CRLF injection, denial of service via complexity attacks, open redirect, arbitrary file read, NTLMv2 hash disclosure, message corruption, and backend routing bypass. These issues affect components such as form-data, brace-expansion, shell-quote, fast-uri, underscore, react-router, @babel/core, js-yaml, launch-editor, websocket-driver, and http-proxy-middleware. The update mitigates these risks by fixing the underlying flaws.
The Chrome extension "Planet Search" with approximately 2 million installs is a search hijacker that routes all user queries through a chain of undisclosed redirects ending at nextgeeker.com, a domain flagged by multiple vendors as a browser hijacker. The extension declares itself as a default search provider but does not disclose the intermediate redirects or the affiliate monetization relationship. The extension's code is minimal and does not reveal this behavior, as the redirect logic is entirely server-side. Users receive standard Google Custom Search results, but their queries transit through third-party domains that may observe search data. The extension is published by FREE VPN PLANET SRL, which also offers other extensions under investigation. The threat was reported to the Chrome Web Store and Google Safe Browsing on 2026-07-25.
The Hubuum client library for Rust versions from 0.0.1 up to but not including 0.6.1 has a vulnerability where authenticated requests may leak the bearer Authorization header outside the configured base path due to how HTTP redirects are handled. The library uses reqwest's default redirect policy, which retains sensitive headers on redirects within the same origin but different paths. An attacker or compromised intermediary influencing a 3xx redirect response could cause the Authorization header to be sent to unintended paths on the same origin. This does not affect cross-origin redirects. The issue is fixed in version 0.6.1 by disabling automatic redirect following. Workarounds include configuring reqwest clients to disable redirects and ensuring the Hubuum origin is not shared with other applications that could redirect requests outside the allowed path prefix.
### Summary
Budibase's central outbound-fetch guard (`fetchWithBlacklist`) prevents SSRF/DNS-rebinding by resolving the target hostname, checking every resolved IP against the blacklist, and **pinning** the connection to the validated IP. The pin is implemented as a Node `http(s).Agent` (`makePinnedAgent`). The fix for CVE-2026-54353 relies on this pin to stop DNS rebinding.
The REST datasource integration (`@budibase/server`) calls `fetchWithBlacklist` but performs the actual request with **undici**'s `fetch`. undici does not support the Node `agent` option — it is silently ignored — and instead uses its own `dispatcher`, which re-resolves the hostname's DNS at connection time. As a result, **the validated/pinned IP is never used on the REST datasource path**, and the DNS-rebinding protection that CVE-2026-54353 added is silently defeated for the single most-used outbound path in Budibase.
An authenticated user who can configure/run a REST datasource (e.g. a builder/tenant) can use a rebinding hostname (public IP during validation, internal IP at connect) to make the server issue arbitrary, full-response HTTP requests to internal-only services — cloud metadata (IAM credential theft), the internal CouchDB/Redis/MinIO, and other internal endpoints — reading and, because REST datasources allow arbitrary method/body, writing or destroying internal data.
### Details
**The guard pins the validated IP via a Node agent** — `packages/backend-core/src/utils/outboundFetch.ts`:
- `resolveSafePinnedIp(url)` resolves the hostname and checks every address against `isBlacklisted`, returning a single `pinnedIp` (lines ~39–53).
- `makePinnedAgent(url, ip)` builds a **Node** `http.Agent`/`https.Agent` whose `lookup` always returns `pinnedIp`, so a node-fetch connection can only reach the validated IP (lines ~55–68).
- `fetchWithBlacklist` passes that agent into the request: `fetchFn(nextUrl, { ...nextRequest, agent: makePinnedAgent(nextUrl, pinnedIp) })` (lines ~186–192). Each redirect hop is re-validated and re-pinned in the loop.
**The REST integration overrides the transport with undici, which ignores `agent`** — `packages/server/src/integrations/rest.ts`:
- `fetch` is imported from **`undici`** (top-of-file import block, ~line 30).
- The request is made by overriding `fetchFn` (lines ~767–793):
```ts
const setDispatcher = (requestInput, requestUrl) => ({
...requestInput,
dispatcher: getDispatcher({ rejectUnauthorized, url: requestUrl }),
})
...
response = await coreUtils.fetchWithBlacklist(url, input, {
fetchFn: async (requestUrl, requestInput) =>
fetch(requestUrl, setDispatcher(requestInput, requestUrl)), // undici.fetch
})
```
The options object reaching `undici.fetch` is `{ ...nextRequest, agent: <pinned Node Agent>, dispatcher: <getDispatcher result> }`. **undici uses `dispatcher` and ignores `agent`.**
**The dispatcher does no IP pinning** — `packages/backend-core/src/utils/fetch.ts`:
- `getDispatcher` → `createDispatcher` → (no proxy env) → `createDirectAgent` = `new Agent({ connect: { rejectUnauthorized } })` (lines ~109–114, ~161–172, ~183). This is a plain undici `Agent` with **no `connect.lookup` / no pin**, so undici resolves the hostname's DNS itself at connect time.
**Net effect (TOCTOU / DNS rebinding):** `fetchWithBlacklist` validates the hostname → safe public IP and builds a pinned Node agent; the REST path then connects via undici, which re-resolves the same hostname independently. With a rebinding domain (TTL 0: public IP during validation, `127.0.0.1` / `169.254.169.254` / internal IP at connect), the request lands on an internal service — exactly the gap CVE-2026-54353's pin was meant to close.
**Scope of impact / why it's REST-specific:** `rest.ts` is the only caller that overrides `fetchFn` with undici. All other outbound sinks (automation `outgoingWebhook`/`n8n`/`make`/`zapier`/`discord`/`slack`, and AI-extract's `processUrlFile`) use the default node-fetch-based `fetchWithBlacklist`, which **does** honor the pinned agent and is **not** affected. REST datasource queries are the most common outbound path, and the response body is returned to the caller (full-response SSRF, not blind).
### PoC
The PoC drives the **real, unmodified** guard code (`outboundFetch.ts` + `fetch.ts`, copied verbatim — sha256 verified) and reproduces the exact `rest.ts` call pattern. Only the `../blacklist` module is stubbed to model the rebinding **input** (validation observes a safe public IP). Requires Node 18+.
```bash
# prerequisite: a Budibase checkout; set BB to its path
export BB=/path/to/budibase
mkdir ssrf-poc && cd ssrf-poc
SRC="$BB/packages/backend-core/src"
# 1) Copy the REAL guard code, verbatim (sha proves no edits)
mkdir -p real/utils real/blacklist
cp "$SRC/utils/outboundFetch.ts" real/utils/
cp "$SRC/utils/fetch.ts" real/utils/
# 2) Scenario stub = the rebinding INPUT: validation sees a safe, non-blacklisted public IP
cat > real/blacklist/index.ts <<