CISA Adds REDCap Flaw to KEV Catalog, Patch Due Today

CISA has added a critical, actively exploited REDCap flaw to its KEV catalog, with a federal patch deadline of today. Here is what CVE-2026-90817 is and how to fix it.

The official seal of the U.S. Cybersecurity and Infrastructure Security Agency, which maintains the Known Exploited Vulnerabilities catalog
Image: CISA.

CISA has added a critical, actively exploited flaw in REDCap — the data-collection platform used by thousands of hospitals, universities and research labs — to its Known Exploited Vulnerabilities catalog, with a federal patch deadline of today, November 21, 2026. CVE-2026-90817 is an unauthenticated remote code execution bug rated 9.8 out of 10 in severity, and attackers have already been caught using it to install a cryptocurrency miner on exposed servers. Here's what the flaw is, who it affects, and the exact versions that fix it.

The short version: CVE-2026-90817 is a 9.8-severity unauthenticated code injection flaw in REDCap's public survey and data-import handling. CISA added it to the KEV catalog on November 18, 2026, with a federal remediation deadline of today. It's fixed in REDCap 16.0.49 LTS, 17.3.10 LTS and 17.4.4 — anything older on the 13.3.0-and-up line is vulnerable. Exploitation has already been observed in the wild, deploying a disguised cryptominer.

What CVE-2026-90817 actually is

REDCap (Research Electronic Data Capture) is a web-based platform built by Vanderbilt University for designing surveys and databases for research studies — the kind of software a hospital uses to run a clinical trial questionnaire or a university lab uses to collect study data. According to the project's own community site, it's deployed at more than 8,500 institutions in 167 countries, running roughly 2.9 million projects for 4.7 million users, with over 59,000 academic papers citing data collected through it.

That scale is exactly what makes a flaw like this one worth taking seriously beyond the CVSS number alone. REDCap instances routinely sit on top of institutional review board-approved research data, which can include identifiable health information gathered under strict consent and privacy frameworks. A university running its own REDCap server for a handful of active studies is a very different risk profile than a cloud SaaS product with a dedicated security team watching it — many institutional deployments are maintained by small IT teams for whom REDCap is one application among dozens, not a primary focus.

The official NVD record for CVE-2026-90817 describes it as an unauthenticated remote code execution flaw "in the survey passthrough routing and Data Import processing logic." An attacker can send a crafted HTTP request that reaches an unintended controller route through a public survey, then supply a manipulated file-path or stream parameter during the import handling, letting them execute arbitrary code on the server. NVD rates it 9.8 on the CVSS 3.1 scale — critical, with the maximum possible impact to confidentiality, integrity and availability — and classifies it under CWE-73 (external control of file name or path) and CWE-94 (code injection). The flaw was originally reported by researcher Ryan Wincey through Securifera's advisory program, the security research firm credited in NVD's reference list alongside CISA's own catalog entry.

A 9.8 is about as severe as the CVSS scale gets, and the reason is the combination of factors involved: no authentication required, low attack complexity, no user interaction needed, and full impact across confidentiality, integrity and availability if it's pulled off. The only thing keeping it off a perfect 10 is that narrow requirement for a valid public survey hash — a real but modest obstacle, since survey links get shared broadly by design, posted in emails, embedded in web pages, and sometimes indexed by search engines, rather than a meaningful barrier against a motivated or automated attacker scanning for exposed links.

How we got from disclosure to today's deadline

The timeline here is longer than the "CISA added it this week" framing suggests. NVD's record shows the vulnerability was published on September 20, 2026, and a third-party tracker covering the disclosure reports that Vanderbilt's REDCap Consortium shipped fixed releases within roughly two days of being notified — a fast turnaround for a critical remote code execution bug. That means the patch has been available for about two months already. What changed on November 18 is that CISA added the flaw to its Known Exploited Vulnerabilities catalog specifically because it found evidence of real-world exploitation, which is a different bar than "a critical flaw exists" — it means attackers are actually using it, not just that they theoretically could.

That gap between "patch available" and "patch applied everywhere" is exactly the window attackers operate in. A fix shipping in September doesn't help a research lab or hospital IT department that didn't prioritize the update, and REDCap's own distribution model — pushing releases out through institutional administrators rather than an auto-update mechanism most end users would see — means adoption of a new version can lag well behind the fix being technically available.

Which versions are vulnerable

NVD's official version data pins down the affected ranges precisely, which matters because REDCap runs multiple release lines in parallel:

Affected lineVulnerable versionsFixed in
Legacy / 16.0 LTS13.3.0 up to (not including) 16.0.4916.0.49 LTS
16.1–17.3 LTS16.1.0 up to (not including) 17.3.1017.3.10 LTS
17.4 Standard17.4.0 up to (not including) 17.4.417.4.4

If you administer a REDCap instance, the practical takeaway is simple: check your exact version number against the ranges above, not just whether you're "on 17.x" or "on an LTS branch." The gap between a vulnerable build and a fixed one can be a handful of point releases.

It's already being actively exploited

This isn't a theoretical risk. CISA only adds a flaw to the KEV catalog when it has evidence of real-world exploitation, and in this case security firm Safe Security has publicly described watching an unknown threat actor abuse the flaw to stand up a transient web shell on compromised REDCap servers. The attacker's payload was an in-memory loader that fetched XMRig, a cryptocurrency-mining tool, disguising it as ordinary image files to slip past casual inspection. Once running, the miner itself was masqueraded as a legitimate system process named syslog-ng, and the attacker set up persistence through a cron job and an added SSH key — standard moves for keeping access to a compromised box even after a reboot or a partial cleanup.

Cryptomining is a comparatively low-stakes outcome compared to what this flaw could enable. Because REDCap servers often hold identifiable research and health data, and because the vulnerability is a full remote code execution bug rather than something narrower, an attacker who already has a working exploit chain for cryptomining could just as easily pivot to data theft on the same servers. There's no indication that's happened here, but the capability is the same regardless of what a given attacker chooses to do with it.

It's also a familiar pattern in how opportunistic attackers monetize newly exploited web vulnerabilities: cryptomining is often the first, lowest-effort use of a freshly compromised server, deployed automatically at scale before a human operator decides whether a particular target is worth a more targeted follow-up. That means a server showing cryptomining activity tied to this flaw isn't necessarily evidence that nothing more serious happened — it may just be evidence of the first stage of a compromise that a fully patched, promptly investigated system would have caught before anything further occurred.

The federal patch deadline is today

CISA's Known Exploited Vulnerabilities catalog entry, added November 18, 2026, sets a remediation due date of November 21 — today — for CVE-2026-90817, alongside flaws in Synology DiskStation Manager and Apache Axis added in the same batch; a fourth flaw in BerriAI's LiteLLM from the same update carries a December 2 deadline instead. The KEV deadline is binding on U.S. federal civilian agencies under binding operational directive, not on private organizations, hospitals or universities — but CISA's own guidance is that everyone running the affected software should treat the same deadline as the practical bar to clear, since the catalog only includes vulnerabilities with confirmed active exploitation.

Pandromeda has covered the other three flaws from this same CISA batch separately: the LiteLLM RCE flaw with its December 2 deadline, and the Apache Axis bug added to CISA's list the same week. If you administer any of these products, this is one batch of advisories to work through together rather than four unrelated incidents.

How to patch and mitigate

The fix is straightforward for any instance that can be updated: upgrade to REDCap 16.0.49 LTS, 17.3.10 LTS, or 17.4.4, depending on which release line you're running. For instances where an immediate upgrade isn't possible, a few mitigations reduce exposure in the meantime:

  • Audit public survey links. Since exploitation requires a valid public survey hash, review which surveys are publicly accessible and whether any can be closed, expired, or switched to an authenticated-access model until you've patched.
  • Restrict network exposure. If your REDCap instance doesn't need to be reachable from the open internet, putting it behind a VPN or IP allowlist removes most opportunistic scanning, even though it won't stop an attacker who already has a valid survey link.
  • Check for signs of compromise. Given the specific indicators from the Safe Security report, look for unexpected cron entries, unrecognized SSH authorized keys, unusual processes named to resemble legitimate system services like syslog-ng, and unexplained CPU load consistent with cryptomining.
  • Separate the web and database tiers. Running REDCap's database server behind its own firewall, isolated from direct internet exposure, limits how far an attacker can move even after compromising the web-facing component.

CISA's own required-action language for this entry points administrators toward its Binding Operational Directive 26-04 guidance on prioritizing security updates by risk, plus a companion set of forensics triage requirements, since this flaw is flagged for mandatory forensic review rather than a routine patch-and-move-on. That distinction matters operationally: CISA is telling federal agencies not just to update REDCap, but to check whether they were already compromised before the patch went in — a step worth taking at any institution running a vulnerable version, given that exploitation has been confirmed since well before this week's KEV addition.

What to do next

If you run REDCap anywhere in your organization — a single lab instance or a shared institutional deployment — confirming the exact version number against the ranges above is the first step, today, not after the next maintenance window. Vanderbilt's REDCap Consortium distributes releases to registered institutional administrators rather than the general public, so the fastest path to the patched build is usually through your institution's existing REDCap administrator rather than a public download. Given that CISA has confirmed active exploitation and the fix has been available since September, there's no reason to wait for a scheduled update cycle on this one.

If your institution also runs Synology DSM, Apache Axis, or LiteLLM — the three other products CISA flagged in the same mid-November KEV batch — treat this as a single coordinated patch sprint rather than four separate fire drills spread across different teams. Checking all four at once is also the more efficient way to confirm whether any single intrusion touched more than one exposed system, rather than investigating each report in isolation and potentially missing a connection between them.

Frequently asked questions

What is CVE-2026-90817?

A 9.8-severity, unauthenticated remote code execution flaw in REDCap's public survey and data-import handling, added to CISA's Known Exploited Vulnerabilities catalog on November 18, 2026.

Which REDCap versions are affected?

REDCap 13.3.0 and later is affected below 16.0.49 LTS, 17.3.10 LTS, or 17.4.4, depending on which release line you run.

Has this vulnerability actually been exploited?

Yes. Security firm Safe Security reported attackers using it to deploy a disguised XMRig cryptocurrency miner on compromised REDCap servers.

What is the patch deadline?

CISA set November 21, 2026 — today — as the federal remediation deadline for U.S. civilian agencies, though the fix has been available since September.

Is REDCap safe to keep using once patched?

Upgrading to a fixed version closes this specific flaw. Routine precautions, like restricting network exposure and auditing public survey links, remain good practice regardless.

What is REDCap used for?

REDCap is research data-collection software built by Vanderbilt University, used by hospitals, universities and labs to run surveys and manage study data for research projects.

Sources

More on REDCap →REDCapCISA KEVCVE-2026-90817VulnerabilityHealthcare IT
Sana Qureshi
Written bySana Qureshi

Sana Qureshi runs the security and privacy desk. She reports on actively exploited vulnerabilities, vendor patches and data breaches, and covers the password managers, VPNs and authentication tools readers use to protect themselves. Her alerts cite vendor advisories, CISA and the CVE record directly.

More from Security & Privacy

See all