Grafana's Prometheus data source passes the resource path supplied by the caller to the configured Prometheus server without restricting it to the supported API endpoints [missing path allowlist]. A user with permission to query the data source can read responses from other endpoints on that server using the data source's credentials.
pgEdge Postgres MCP Server enables SQL queries against PostgreSQL databases through Model Context Protocol compatible clients. Prior to 1.1.0, custom pl-do tools interpolated JSON arguments into a PostgreSQL DO block delimited by fixed dollar-quote tags. An argument containing $mcp_custom_tool$ could terminate the block and inject arbitrary SQL statements, while $mcp_args$ could inject procedural code into a plpgsql block. Because the statement used the simple query protocol, injected text could include additional statements and transaction control, bypassing read_only and accessing data or modifying resources available to the connecting role. The issue requires at least one pl-do custom tool; sql and pl-func tools pass arguments as bind parameters and are not affected by this injection mechanism. This issue is fixed in 1.1.0.
pgEdge Postgres MCP Server enables SQL queries against PostgreSQL databases through Model Context Protocol compatible clients. Prior to 1.1.0, query_database on a read_only connection could send unrecognized statements through pgx Exec without bind parameters, selecting PostgreSQL's simple query protocol and allowing multiple semicolon-separated statements in one request. A caller could combine transaction or session state changes with database writes, and pooled session state could leave later calls writable, when the connecting role held write privileges. The flaw affected deployments relying on read_only as a safety boundary, could compromise integrity and availability, and did not add confidentiality impact beyond the reads already permitted by the connection. This issue is fixed in 1.1.0.
A vulnerability in PostgreSQL prior to versions 18.7, 17.12, 16.16, 15.20, and 14.25 allows an object creator to execute arbitrary SQL as the caller of a security-restricted operation. This occurs due to improper protection of an alternate path involving domain, enum, or range types created in the temporary schema pg_temp. The exploit requires the victim session to run a query with an unqualified type name, which common security-restricted operations like autovacuum do not perform.
An authenticated remote code execution vulnerability exists in Zohocorp ManageEngine products including AssetExplorer, ServiceDesk Plus, ServiceDesk Plus MSP, and SupportCenter Plus. The flaw affects versions below 7860 for AssetExplorer, below 15290 for ServiceDesk Plus, and below 15150 for both ServiceDesk Plus MSP and SupportCenter Plus. Exploitation requires an authenticated user with the Create Query Reports permission to craft malicious database queries that execute arbitrary commands on the server.
Blind SQL injection vulnerability in the public endpoint `public/webhook/receiver.php` of the WhatsApp module for Dolibarr 1.9. The phone number value derived from the `data.key.remoteJid` parameter, received via a JSON request, is incorporated into an SQL query without escaping or parameterization. With the WHATSAPP_WEBHOOK_ALLOW option enabled, an unauthenticated remote attacker could manipulate this parameter to alter the query. Exploitation using a timing-based technique could allow an attacker to extract information from the database and cause service disruption.
The Wikidata Lookup expansion module (wiki.py) in MISP misp-modules contained a SPARQL injection vulnerability. The handler function constructed a SPARQL query by directly concatenating the user-supplied text value into the query string without any escaping or parameterization: 'SELECT ?item WHERE { ?item rdfs:label"' + request.get("text") + '" @en }'. An authenticated MISP user with access to the module could supply a crafted text attribute value containing SPARQL syntax (quotes, closing braces, UNION, SERVICE, LIMIT, etc.) to alter the structure of the query sent to the Wikidata SPARQL endpoint. This could allow an attacker to modify the query logic, extract unintended data, or cause the query to fail in ways that expose internal error details. Additionally, the old error handling path formatted raw exception objects into the response string ('wikidata API not accessible {}'.format(e)), potentially leaking internal information such as proxy credentials, query contents, or response bodies. The module also used a spoofed browser user-agent string and the deprecated Wikidata endpoint. The vulnerability required authenticated access to the MISP instance and the ability to invoke the wiki expansion module with a text attribute. The impact is limited to manipulation of the enrichment query and potential disclosure of internal error details; it does not directly compromise the MISP server or its stored data.
A blind SQL injection vulnerability exists in Revive Adserver version 6.0.8 and earlier. This flaw allows an authenticated manager, restricted to one agency, to extract arbitrary database values via the banner-zone.php and campaign-zone.php administration pages. The vulnerability arises from inconsistent request-input precedence in permission checks and database queries, enabling an attacker to pass validation with an authorized identifier while injecting a different value into the SQL query. Extracted data can include sensitive information such as password hashes and active session identifiers. Reusing an unexpired administrator session can lead to full administrative access, enabling configuration changes and disruption of ad delivery. The issue is fixed in version 6.0.9 by correcting input handling and excluding cookies from input helpers.
HHVM versions prior to commit 23c061f4e804a51badd5f85227e65185651253b5 contain a vulnerability in request variable parsing. Specifically, when handling query-string and form parameters, HHVM did not verify that appending a nested array succeeded before further processing. This flaw allows a remote, unauthenticated attacker to send crafted GET or POST requests that cause the HHVM server process to crash before application code execution, resulting in denial of service.
CVE-2026-64861 is a resource exhaustion vulnerability in the Probo GRC platform prior to version 0.222.0. The GraphQL endpoint does not limit alias count, query depth, or query complexity, allowing attackers to craft requests with thousands of aliased resolver calls. This can cause excessive server resource consumption and degrade service availability. The issue is fixed in version 0.222.0.
The Joinchat – Enhanced "click to chat" plugin for WordPress is vulnerable to Reflected Cross-Site Scripting via '{HREF}' Dynamic Variable via Query Parameter Name in all versions up to, and including, 6.3.1 due to insufficient input sanitization and output escaping. This makes it possible for unauthenticated attackers to inject arbitrary web scripts in pages that execute if they can successfully trick a user into performing an action such as clicking on a link. Exploitation requires that an administrator has configured a telephone number and included the {HREF} dynamic variable in the Call-to-Action text.
A vulnerability in MongoDB Server's user management commands causes incomplete cleanup of role authorization data when roles are dropped. This flaw results from incorrect internal query predicates that remove role definitions but leave stale references in user and role documents. If a role is recreated with the same name but different privileges, users previously associated with the role may inherit new privileges silently, without explicit grants or audit events. Exploitation requires authenticated database administrative access.
CVE-2026-125327 is a high-severity vulnerability in pgEdge pgedge-rag-server versions prior to 2.0.0. The unauthenticated query endpoint reads the entire filtered corpus without a LIMIT and rebuilds an in-memory BM25 index on every request, causing excessive CPU and database resource consumption. This can lead to service resource exhaustion via inexpensive remote queries. The issue is fixed in version 2.0.0.
CVE-2026-125325 is a race condition vulnerability in pgEdge pgedge-rag-server versions prior to 2.0.0. The BM25 keyword index used by the unauthenticated query endpoint is shared across requests and its operations are locked separately, allowing concurrent requests to access data from other tenants. This can lead to unauthorized disclosure of documents belonging to different customers. The issue is fixed in version 2.0.0.
CVE-2026-92684 is a high-severity vulnerability in Neo4j's @neo4j/graphql library versions 5.0.0 before 5.12.17 and 7.0.0 before 7.6.8. It involves missing authentication enforcement on critical aggregation query functions, allowing unauthenticated remote attackers to read protected data. Specifically, field-level authentication is not applied to string aggregates, and connection-root aggregates bypass aggregate authentication requirements, exposing sensitive data via GraphQL queries.
Multiple security vulnerabilities were discovered in the Linux kernel's io_uring subsystem and other components. These include memory leaks, improper validation of query request sizes, incorrect handling of canceled task work, and improper tracking of buffer consumption. These flaws could allow local attackers to cause denial of service conditions such as memory exhaustion, system hangs, or crashes, and remote attackers to corrupt receive buffers leading to denial of service or other impacts. The update also addresses numerous other security issues across a wide range of kernel subsystems and drivers.
CVE-2026-115934 is a critical SQL injection vulnerability in CraftCMS affecting versions 4.0.0 up to but not including 4.18.9, and 5.0.0 up to but not including 5.11.2. It allows unauthenticated attackers to perform stacked SQL queries via the craft.search front-end search feature on MySQL/MariaDB databases. This can lead to unauthorized read and write access to the application database, including privilege escalation to administrator by modifying user records. The vulnerability is fixed in versions 4.18.9 and 5.11.2. Temporary mitigation includes blocking front-end requests containing the attribute parameter in the query string.