← All updates

Security / News explained

REDCap’s exploitation warning starts with a public survey, not an account

CISA added REDCap’s CVE-2026-90817 to its exploited-vulnerability catalog. The flaw crosses from a public survey into import processing and requires a valid public survey hash.

By Zyel · Published · 4 min read

CISA added REDCap vulnerability CVE-2026-90817 to its Known Exploited Vulnerabilities catalog on November 18, 2026. The entry concerns code execution reached through a public survey and unintended import processing. Exploitation requires knowledge of a valid public survey hash, but does not require an authenticated REDCap account. [1]

This is a new exploitation listing for older disclosed research. Securifera’s CVE record was published on September 20, and its detailed public explanation appeared on September 27. Keeping those stages separate avoids turning an existing vulnerability into a newly discovered November flaw. [2] [3]

The public CVE description traces affected code back to REDCap 13.3.0. Its remediation names 16.0.49 LTS, 17.3.10 LTS and 17.4.4 Standard Release, as applicable. Those are branch-specific fixed releases; the description should not be read as declaring later patched versions vulnerable simply because their numbers are higher. [2]

A public link crossed several internal boundaries

The original research describes inconsistent URL decoding in survey passthrough checks. An allowed file and the restriction on its permitted route were evaluated using different representations of the same input. That let a public survey request reach an unintended internal controller, even though the file itself passed an allowlist. [3]

The researcher then traced rights being calculated before the survey identity was established. Import processing trusted a supplied resource as a path or stream, and a cleanup write could affect a compiled template cache that the application later executed. These were connected application behaviors, rather than one missing password check. [3]

A useful way to read that chain is to ask what each boundary was supposed to mean. A survey link should establish access to the survey. An approved dispatcher file should not imply access to every action it can dispatch. A resource supplied for import should not automatically become a destination for executable application content.

Those are our engineering conclusions from the reported mechanism. They do not require publishing a working exploit to be useful. They explain why a team can have a legitimate public feature and an apparently protected administration area while still needing to review the transitions between them.

The survey hash is a prerequisite with limits

The valid-hash condition deserves to stay in the description. The CVE record says it is required, while also identifying the attack as unauthenticated. An account requirement and knowledge of a public survey identifier are different conditions. Neither should be silently substituted for the other when an owner assesses exposure. [2]

Consider a hypothetical organization that shares an active survey link with external participants. The intended interaction is answering that survey. A successful demonstration that those participants cannot use the normal administrator login would not establish that every route reachable through the public survey is appropriately restricted.

Conversely, knowing that the software contains the flaw does not establish which survey links were distributed, which installation was reachable, or whether anybody used the vulnerable path. Those are environment-specific questions. The catalog listing is a reason to investigate them promptly, rather than permission to invent their answers.

CISA’s entry records a November 21, 2026 due date and forensic triage “Yes.” Its ransomware-campaign field is “Unknown.” Those are the recorded catalog fields. Unknown ransomware use is not evidence of absence, and the triage flag is not a diagnosis of a particular organization’s server. [1]

An update and a historical assessment do different work

The original researcher recommends upgrading and describes restricting public surveys and relevant survey-context routes as temporary containment when an immediate update is unavailable. That containment advice is not presented as a permanent substitute for the security fixes, or as a complete assessment of earlier activity. [3]

For the hypothetical survey owner, the release branch, the team maintaining the installation and the appropriate vendor support channel determine how an update becomes a supported change. A list of three fixed versions is useful evidence, but it cannot choose an organization’s maintenance path without that context.

Historical review has a separate purpose. An installed fixed release can address the known defect going forward while leaving questions about past access unanswered. Similarly, an ordinary successful survey response demonstrates useful service availability without reconstructing every request that reached the application before the update.

REDCap’s warning is therefore about a chain of authority: public survey access, internal routing, import resources and executable cache content. The new exploitation listing raises the priority of closing that chain’s vulnerable path. A defensible response keeps the fixed release, the public-link prerequisite and the actual evidence about the installation together.

Sources + further reading

Source dates describe the original material. Publication above is the date this Zyel article went live.

  1. CISA Known Exploited Vulnerabilities catalogCISA · Checked Nov 19, 2026
  2. CVE-2026-90817 Securifera CNA recordSecurifera · Checked Nov 19, 2026
  3. Original REDCap routing and import researchSecurifera · Checked Nov 19, 2026

Reader discussion

What would you ask next?

Questions, useful corrections and thoughtful disagreement are welcome. Comments are reviewed before they appear.

No reader comments yet.

10–2,000 characters. Plain text. Leave out passwords, private project data and personal contact details. A blank display name appears as Anonymous.

Complete the CAPTCHA before sending. Verification runs on your device.