Waku CVE-2026-49455
MEDIUMSeverity by source
AV:N/AC:L/PR:N/UI:R/S:U/C:N/I:H/A:N
Network CSRF requiring victim page visit (UI:R); CORS-safelisted POST eliminates preflight complexity (AC:L); no attacker credentials needed (PR:N); SOP blocks response reads (C:N) but server action executes fully (I:H).
Primary rating from GitHub Advisory.
CVSS VectorGitHub Advisory
CVSS:3.1/AV:N/AC:L/PR:N/UI:R/S:U/C:N/I:H/A:N
Lifecycle Timeline
2DescriptionGitHub Advisory
Summary
Waku's RSC request dispatcher invokes server actions without validating the request's Origin (or Sec-Fetch-Site) header. A cross-origin web attacker can therefore cause a victim browser to issue an authenticated POST to a registered server action endpoint using a CORS-safelisted content type (text/plain), which does not trigger a preflight. Any state-mutating server action that the application exposes via 'use server' can be invoked with the victim's cookies attached. A working proof-of-concept demonstrates the vulnerability against waku 1.0.0-beta.0 dev server: a cross-origin POST with Content-Type: text/plain invokes a registered 'use server' action and returns HTTP 200 with an RSC stream response. The same defect affects the progressive-enhancement (no-JavaScript) server action path: a cross-origin HTML form auto-submitting multipart/form-data reaches the dispatch through a second unguarded branch of the request handler, dynamically confirmed on 2026-05-17. Both branches were confirmed exploitable from opaque-origin contexts (sandboxed iframes, file:// navigation, browser extension pages), which send Origin: null - a value no Origin guard exists to reject. This is the same vulnerability class previously disclosed for Next.js Server Actions (GHSA-mq59-m269-xvcx); waku's implementation is broader in that no Origin check exists at all in the default request handler.
Root cause
In packages/waku/src/lib/utils/request.ts:
// line 29
if (pathname.startsWith(rscPathPrefix)) {
rscPath = decodeRscPath(pathname.slice(rscPathPrefix.length));
const actionId = decodeFuncId(rscPath);
if (actionId) {
const body = await getActionBody(req);
const args = await decodeReply(body, { temporaryReferences });
const action = await loadServerAction(actionId);
// ...action is invoked with attacker-supplied args
}
}The sibling else if (req.method === 'POST') branch (line 61) does enforce a method check, but only for non-RSC paths. The RSC dispatch branch has no equivalent guard. For comparison, frameworks in the same category (Next.js, Remix) compare the request's Origin header against the configured host before invoking a server action.
Trigger (one-line summary)
A cross-origin POST with Content-Type: text/plain to /<rscBase>/<encoded-action-path> reaches loadServerAction and invokes a registered 'use server' action with the victim's browser-attached cookies.
Affected entry surfaces
- All HTTP adapters exposed by waku (
packages/waku/src/adapters/node.ts,cloudflare.ts,vercel.ts,edge.ts) - each chains through the samerequest.ts:getInputdispatcher. - Any server action declared with
'use server'and reachable from the Vite module graph (i.e., normal application code).
Suggested fix outline
Add an Origin (and ideally Sec-Fetch-Site) validation step in the RSC dispatch branch of getInput, gated against the configured base URL host, with an opt-in allowedOrigins configuration option for legitimate cross-origin scenarios. Exact patch sketches will be provided in the follow-up comment.
Disclosure
| Field | Value |
|---|---|
| Reporter | j0hndo (dohyun4466@gmail.com) |
| Discovery date | 2026-05-15 |
| Embargo | 90 days from acknowledgement (operator open to extension on request) |
| Comparable precedent | Next.js GHSA-mq59-m269-xvcx ("null origin can bypass Server Actions CSRF checks"), CVSS 5.3, fixed Next.js 16.1.7 |
| References | OWASP CSRF cheat sheet; Fetch spec § CORS-safelisted request-header (text/plain). |
Workarounds (publish-safe)
Until a framework-level fix is released, operators can mitigate by placing an Origin-validating reverse proxy or middleware in front of waku that rejects requests whose Origin header does not match the application's host on POST requests to the configured rscBase prefix. Note that Vite's server.allowedHosts guard (which rejects unrecognized Host headers with HTTP 403) applies only to waku dev; production deployments using waku build && waku start do not have an equivalent guard and are fully exposed without an external mitigation layer.
AnalysisAI
Cross-Site Request Forgery in Waku's RSC request dispatcher allows a remote attacker to invoke any state-mutating 'use server' server action with a victim's session cookies by exploiting the complete absence of Origin header validation in the core request handler at packages/waku/src/lib/utils/request.ts. Both the RSC path prefix branch (exploitable via Content-Type: text/plain, no preflight) and the progressive-enhancement HTML form branch (exploitable via multipart/form-data) are unguarded, with opaque-origin contexts such as sandboxed iframes and browser extension pages additionally exploitable via Origin: null. …
Unlock full vulnerability intelligence
- Risk assessment & exploitation conditions
- Attack chain visualization
- Remediation with exact patch versions
- Threat intelligence from 22 sources
- Personal watchlist & email alerts
Free forever · No credit card required
Attack ChainAIDerived
Hypothetical attack flow derived from CVE metadata
Vulnerability AssessmentAI
| Exploitation | The victim must be authenticated to a Waku application that exposes at least one 'use server' server action and must visit an attacker-controlled web page, or be subjected to an opaque-origin context (sandboxed iframe injected by a third-party script, file:// page, or browser extension) that can auto-issue a cross-origin POST. … Additional conditions and limiting factors are described in the full assessment. |
| Risk Assessment | The CVSS 6.5 score with vector CVSS:3.1/AV:N/AC:L/PR:N/UI:R/S:U/C:N/I:H/A:N correctly captures the threat model: exploitation requires victim user interaction (UI:R - the victim must visit an attacker-controlled page), but imposes no authentication burden on the attacker and no configuration complexity (AC:L). … Full risk analysis with EPSS, KEV, and SSVC signal comparison available after sign-in. |
| Exploit Scenario | An attacker hosts a web page containing a JavaScript fetch call or auto-submitting HTML form that POSTs to the target waku application's RSC endpoint (e.g., /_rsc/<encoded-action-id>) with Content-Type: text/plain or multipart/form-data. When an authenticated victim visits the attacker's page, the browser attaches the victim's session cookies without triggering a CORS preflight; the waku dispatcher decodes the action ID and invokes the targeted server action with attacker-supplied arguments, completing the state mutation (e.g., account modification, data deletion, or privilege change) as the victim. … |
| Remediation | No vendor-released patch has been identified at time of analysis. … Detailed patch versions, workarounds, and compensating controls in full report. |
Threat intelligence, references, and detailed analysis are available after sign-in.
Same weakness CWE-352 – Cross-Site Request Forgery (CSRF)
View allShare
External POC / Exploit Code
Leaving vuln.today
GHSA-75w3-gmqx-993q