ZeroHour

CVE-2026-82291

niche

CORS Origin-Reflection Flaw in HeyForm Exposes Authenticated GraphQL Data

CVSS 3.1
8.1 high
EPSS
<1%p22
Published
()
Modified
AI analysis

HeyForm before 3.0.0-rc.8 reflects whatever Origin header a request supplies in its CORS responses while still allowing credentials, so any website is treated as a trusted cross-origin source. An attacker hosts a malicious page, and when a logged-in HeyForm user visits it, the page can issue authenticated GraphQL queries cross-origin with the user's credentials. This lets the attacker read workspaces, projects, forms, submissions, and respondent data, or change the victim's account settings. Any self-hosted HeyForm deployment running an affected version is exposed whenever its logged-in users can be lured to attacker-controlled pages. There is currently no public proof-of-concept, the EPSS score is low (0.3% probability of exploitation in 30 days), and the flaw is not in CISA KEV, so no exploitation is known.

What to do: Upgrade HeyForm to 3.0.0-rc.8 or later. As an interim mitigation, restrict the CORS configuration to a fixed allowlist of known origins instead of reflecting arbitrary Origin headers with credentials enabled. Review recent account-setting changes and GraphQL access to workspaces, forms, and submissions for signs of unauthorized access.

Affected
HeyFormbefore 3.0.0-rc.8
Estimated exposure
nichelikely low thousands of self-hosted instances at most (open-source, self-hosted form builder; no public install counts) — HeyForm is an open-source, self-hosted form builder with a modest community footprint and no public active-install or internet-scan counts, so the estimate reflects typical small-scale self-hosting adoption rather than measured deployment…

Order-of-magnitude estimate by the model from install counts, market share and public scan data it knows; verify before quoting.

Description

HeyForm before 3.0.0-rc.8 reflects the request Origin header in CORS responses while allowing credentials, enabling cross-origin requests with authentication. Attackers can execute authenticated GraphQL queries from malicious pages visited by logged-in users to access workspaces, projects, forms, submissions, and respondent data, or modify account settings.

Weakness
CWE-942
Vector
CVSS:3.1/AV:N/AC:L/PR:N/UI:R/S:U/C:H/I:H/A:N

In the news

No ingested article mentions this CVE yet.