Disclaimer: This content was generated using AI. While we strive for accuracy, please verify critical information with official sources.
Sources: National Vulnerability Database, CISA KEV, Exploit-DB
CVE-2026-73976 Overview
CVE-2026-73976 is a SPARQL injection vulnerability in djehuty, a research data repository system developed by 4TU.ResearchData. Versions prior to 26.3.2 accept unsanitized input in three search and listing parameters that are concatenated into SPARQL SELECT queries. An unauthenticated attacker can inject arbitrary SPARQL clauses to read data from graphs outside the request scope or submit expensive queries that exhaust backend resources. The flaw is tracked under CWE-943: Improper Neutralization of Special Elements in Data Query Logic.
Critical Impact
Unauthenticated attackers can exfiltrate drafts, private records, and internal triples from the RDF store, and can degrade service availability by submitting expensive SPARQL queries.
Affected Products
- 4TU.ResearchData djehuty, all versions prior to 26.3.2
- Deployments exposing the search and listing endpoints to untrusted networks
- RDF stores backing djehuty instances that contain non-public graphs
Discovery Timeline
- 2026-10-01 - CVE-2026-73976 published to the National Vulnerability Database
- 2026-10-02 - Last updated in NVD database
Technical Details for CVE-2026-73976
Vulnerability Analysis
djehuty exposes search and listing endpoints that build SPARQL queries from user-controlled parameters. Three of these parameters are concatenated into the query text without sanitization or parameterization. An attacker can close the intended clause and append their own SPARQL fragments, including UNION patterns that pull triples from graphs the request was never scoped to.
Because the affected queries are read-only SELECT statements, the vulnerability does not modify the triple store. It does, however, break the data-partitioning assumptions that djehuty relies on to separate draft, private, and internal records from published content. Attackers can also submit queries engineered to be computationally expensive, tying up the SPARQL backend and web workers.
No authentication, privileges, or user interaction are required to reach the vulnerable code path.
Root Cause
The root cause is string concatenation of untrusted input into SPARQL queries, classified as CWE-943. The affected handlers trust the three injectable parameters as query fragments rather than treating them as literal values. SPARQL has no equivalent of prepared-statement binding by default, so missing input neutralization directly becomes query injection.
Attack Vector
An attacker sends crafted HTTP requests to djehuty search or listing endpoints with malicious values in the vulnerable parameters. By injecting UNION { GRAPH ?g { ?s ?p ?o } } style patterns, the attacker enumerates triples from arbitrary named graphs, including those storing drafts and internal metadata. For denial of service, the attacker injects unbounded joins or regex-heavy filters that force the backend to materialize large intermediate result sets.
For technical details on the injection points and corrected queries, see the GitHub Security Advisory GHSA-7gp2-rxw9-vw6p.
Detection Methods for CVE-2026-73976
Indicators of Compromise
- HTTP requests to djehuty search or listing endpoints containing SPARQL keywords such as UNION, GRAPH, SELECT, OPTIONAL, or FILTER inside query parameters
- Unusually large response payloads from search endpoints that normally return bounded result sets
- SPARQL backend logs showing queries referencing graph IRIs that are not part of the public API surface
- Spikes in query execution time, worker saturation, or 5xx responses from listing endpoints
Detection Strategies
- Inspect web server and reverse proxy logs for SPARQL syntax tokens appearing in query string or form parameters bound for djehuty routes
- Enable SPARQL endpoint query logging and alert on queries that reference named graphs outside the published allow-list
- Baseline normal query shapes for djehuty search endpoints and flag deviations in clause count, join depth, or execution time
Monitoring Recommendations
- Forward djehuty application logs and SPARQL backend logs to a centralized analytics platform for correlation
- Monitor response size and latency per endpoint to detect data exfiltration and resource-exhaustion patterns
- Alert on repeated requests from a single source that return anomalous volumes of triples or trigger backend timeouts
How to Mitigate CVE-2026-73976
Immediate Actions Required
- Upgrade djehuty to version 26.3.2 or later, which contains the fix for the three injectable parameters
- Audit SPARQL backend logs for prior queries containing injected UNION or GRAPH clauses targeting non-public graphs
- Rotate any secrets or tokens that may have been stored as RDF triples in graphs reachable through the injection
Patch Information
The maintainers released the fix in the djehuty v26.3.2 release. The accompanying GitHub Security Advisory GHSA-7gp2-rxw9-vw6p documents the affected parameters and the remediation.
Workarounds
- Restrict network access to djehuty search and listing endpoints until the upgrade is applied
- Deploy a web application firewall rule that rejects requests whose search parameters contain SPARQL keywords or unexpected punctuation such as }, {, or ;
- Configure the SPARQL backend to enforce per-query timeouts and result-size limits to blunt denial-of-service attempts
- Restrict the service account used by djehuty so it can only query the graphs required for public listings
Disclaimer: This content was generated using AI. While we strive for accuracy, please verify critical information with official sources.
Sources: National Vulnerability Database, CISA KEV, Exploit-DB

