Attack ChainAIDerived
Hypothetical attack flow derived from CVE metadata – not a confirmed exploit chain
-
AccessReach public search/listing endpoint
-
ExploitInject SPARQL via query parameter
-
ExecutionUNION in out-of-scope graph triples
-
ImpactExfiltrate private/draft data or exhaust SPARQL backend
Vulnerability Assessment
| Weakness |
CWE
OWASP Top 10:2025
A05:2025 - Injection
ASVS 5.0
V1 - Encoding and Sanitization Query language parameterization
|
| Exploitation Conditions | No special conditions beyond network reachability of the affected feature - exploitation targets djehuty's unauthenticated search/listing functionality, where SPARQL can be injected through three separate request parameters against default deployments prior to version 26.3.2; no account, no credentials, and no user interaction are required. Limiting factors are modest: the injected queries are read-only SELECT operations, so no data can be written or modified in the RDF store, and realistic impact is bounded to (a) reading triples from graphs outside the request's intended scope (private/draft/internal data) and (b) resource-exhaustion DoS against the SPARQL backend and web workers. The only real deployment constraint is that the instance must be running a version earlier than 26.3.2 with its search/listing endpoints exposed. |
| Impact | This is a genuine, actionable priority rather than a CVSS-inflated non-issue, though the available signals conflict on one key point. The official CVSS 4.0 vector (AV:N/AC:L/AT:N/PR:L/UI:N) scores 7.1 (High) and encodes network-reachable, low-complexity exploitation, but marks it as requiring low privileges (PR:L); the vendor description flatly contradicts this, stating 'An unauthenticated attacker' and 'No account or user interaction is required,' which argues for PR:N and a higher effective score. The impact mapping in the official vector is also questionable: it records VI:H (integrity) with VC:N, yet the described harm is cross-graph data exfiltration (a confidentiality impact) and DoS (an availability impact), with no write to the store - so the vector appears to mislabel the impact dimension. No EPSS score, CISA KEV listing, or public POC was provided in the intelligence, so there is no evidence of active exploitation or mass scanning; exploitation status must be treated as theoretical for now. Net assessment: an unauthenticated SPARQL injection in a network-facing search endpoint is inherently easy to exploit and should be patched promptly, but real-world urgency is tempered by the product's niche academic footprint and the absence of any exploitation telemetry. |
| Exploit ScenarioDerived | An unauthenticated attacker sends a crafted request to djehuty's public search or listing endpoint, injecting SPARQL fragments into one of three vulnerable parameters. Because the backend queries are SELECT (read) operations, the attacker UNIONs in triples from graphs the request was never scoped to, pulling back private, draft, or internal RDF data; alternatively they submit expensive or malformed queries to tie up the SPARQL backend and web workers. No public exploit code was identified at time of analysis, but the technique is standard query-injection and requires no authentication or user interaction. |
Get an e-mail when this CVE is exploited, patched or re-scored. Free account.
Alert meAnalysisAIv1 – Oct 01, 2026
Unauthenticated SPARQL injection in djehuty, the 4TU.ResearchData research data repository, before version 26.3.2 lets remote attackers manipulate the SELECT queries behind the search and listing endpoints through three separate request parameters, enabling cross-graph data exfiltration and resource-exhaustion denial of service. Because no account, credentials, or user interaction are required, any instance exposing its search/listing functionality to the network is at risk; the injected queries are read-only, so no data can be written or altered, but triples held in graphs outside the request's intended scope (drafts, private, or internal RDF data) can be pulled in via UNION-style constructs, and deliberately expensive or malformed queries can tie up the SPARQL backend and web workers. A vendor-released patch exists in 26.3.2 and has been published as advisory GHSA-7gp2-rxw9-vw6p, and no public exploit code or confirmed active exploitation was identified at time of analysis.
RemediationAI
Vendor-released patch: 26.3.2 - upgrade from any earlier version to djehuty 26.3.2 (https://github.com/4TUResearchData/djehuty/releases/tag/v26.3.2), which applies the search-order allow-list fix (commit 897c915) that removes the injectable query structure; consult advisory GHSA-7gp2-rxw9-vw6p for the vendor's guidance. If the upgrade cannot be scheduled immediately, compensating controls should target the three injected parameters on the search and listing endpoints: place those routes behind an authenticating reverse proxy or gateway so only trusted clients reach them (trade-off: this breaks the intended public browsing experience for a public data repository), or, if they must stay public, add strict rate limiting and a short server-side SPARQL query timeout plus result-size caps at the proxy layer (trade-off: legitimate heavy queries will start failing and time out, and timeouts must be tuned against normal traffic). Tightening the SPARQL backend's credentials so the application can only read the specific named graphs it needs - with drafts, private, and internal graphs excluded - limits cross-graph exfiltration even if injection succeeds (trade-off: any legitimate feature that joins across those graphs will break). Blocking or normalizing UNION, GRAPH, and SERVICE keywords and unusual ORDER BY clauses at a WAF is a stopgap only (trade-off: trivially bypassable via encoding and prone to false positives against legitimate queries), and it should be paired with log monitoring for anomalous query lengths, UNION keywords, or sudden 5xx/timeout spikes from the search endpoints. No vendor-released patch beyond 26.3.2 is required, and no public exploit code was identified at time of analysis, so prioritization should be driven by whether search/listing endpoints are internet-exposed.
Affected VersionsENISA
Affected Packages
Source: OSV (Open Source Vulnerabilities)
Affects your stack
Affects 1 package in euvd
Checking this CVE against your own stacks and servers is part of the Basic plan and up.
Technical ContextAI · show details
djehuty is a Python-based research data repository (CPE cpe:2.3:a:4turesearchdata:djehuty:*:*:*:*:*:*:*:*) that stores metadata and related records in an RDF triplestore and exposes search and listing features backed by SPARQL queries. The weakness is CWE-943, improper neutralization of special elements in data query logic - the classic NoSQL/query-language injection class applied to SPARQL rather than SQL, and the reason the issue is tagged as a NoSQL-injection variant. Rather than parameterizing or validating untrusted input before it is concatenated into query text, three distinct request parameters flow into the search/listing query strings, allowing an attacker to append or restructure SPARQL clauses. Since the affected operations are read-only SELECT statements, the impact is confined to disclosure and availability: injected UNION / GRAPH constructs can reach named graphs the request was never scoped to (drafts, private, or internal datasets), and unbounded or pathological query patterns can consume backend and worker capacity. The v26.3.2 release notes corroborate the fix path, listing a change that restricts search order to an allow-list (commit 897c915) alongside an unrelated escaping fix for the session name field (commit f9f3e47); the allow-listing of query structure is the mitigation that closes the injection surface. Note that the vendor-published CVSS 4.0 vector differs from our independent assessment, which scores the flaw as network-reachable, low complexity, no privileges (PR:N), no user interaction, with high confidentiality and high availability impact; the described exploitation path targets unauthenticated endpoints, and the assessment vector is the one reflected here.
Know when Djehuty gets its next vulnerability
- Watch products and get an email when something new is published
- Check this CVE against your own stack and servers
- Exploit and PoC references, threat intelligence from 22 sources
No credit card · 7-day full trial
Severity by source
CVSS:4.0/AV:N/AC:L/AT:N/PR:L/UI:N/VC:N/VI:H/VA:N/SC:N/SI:N/SA:N
Vendor states unauthenticated, no interaction (PR:N/UI:N) over the network at low complexity; read-only injection yields high confidentiality (cross-graph exfiltration) and high availability (backend DoS) but no integrity impact.
Primary rating from Vendor (GitHub_M).
CVSS VectorVendor: GitHub_M
Lifecycle Timeline
2DescriptionCVE.org
djehuty is a research data repository system developed by 4TU.ResearchData. Prior to version 26.3.2, An unauthenticated attacker can inject SPARQL into the search/listing queries through three separate parameters. Because the affected queries are read (SELECT) queries, this does not write to the store, but it allows: Cross-graph data exfiltration - e.g. UNION-ing in triples from graphs the request was never scoped to (drafts/private/internal data held in the RDF store); denial of service - expensive or malformed queries that tie up the SPARQL backend / web workers. No account or user interaction is required. This issue has been patched in version 26.3.2.
Technical POC Analysis
CWE-943 (Improper Neutralization of Special Elements in Data Query Logic).
Full technical POC analysis - root cause, step-by-step exploitation and patch breakdown - is part of the Pro and Business plans.
Executive ViewAI
The 4TU.ResearchData repository platform (djehuty) is exposed to a query-manipulation flaw in its search and listing endpoints that can pull research data from graphs outside a request's intended scope - including draft, private, and internal RDF content - and can also degrade or take down repository availability through costly queries. Any network-reachable deployment running a version prior to 26.3.2 is affected, so institutional research data governed by confidentiality or embargo obligations may be at risk, alongside service availability for researchers and partners who depend on the repository. Leadership should treat this as a data-governance and availability exposure rather than a data-integrity issue, since the injected queries are read-only and cannot alter or write records.
Apply vendor patch.
Threat Intelligence Overview
How this decision was reached
The engine reads each signal below and applies a fixed list of rules from top to bottom. The first rule that matches sets the recommended action – like a statement, every line is shown.
What every action level means
SSVC AssessmentENISA
Priority Score
Attack Technique
NoSQL injection exploits databases that don't use traditional SQL syntax-like MongoDB, CouchDB, or Redis-by manipulating queries through unsanitized user input.
Denial of Service attacks render applications or systems unavailable by overwhelming resources or triggering failure conditions.
EUVD-2026-90967
NVD
CVE.org
EPSS
Vulnrichment
GitHub
Exploit-DB