CVE Watch11 ago 2026

CVE-2026-48774

ProxySQL is a proxy for MySQL and its forks, as well as PostgreSQL. In versions 3.0.0 through 3.0.8, ProxySQL's GenAI/MCP `run_sql_readonly`

CVSS

7.5

Alto

EPSS

0.3%

p19

KEV

—

Exploit Today

0-100

Publicado: 19 jun 2026 · Última mod.: 11 ago 2026 · CWE-20

EPSS · 30d

0.3%EPSS · 30 días0.4%

2026-08-062026-09-02

Descripción técnica

ProxySQL is a proxy for MySQL and its forks, as well as PostgreSQL. In versions 3.0.0 through 3.0.8, ProxySQL's GenAI/MCP `run_sql_readonly` tool violates its documented read-only contract for MySQL targets. The tool validates only the full input string with a substring blacklist and first-keyword allowlist, but then executes the entire SQL string on a backend connection created with `CLIENT_MULTI_STATEMENTS`. As a result, a caller can submit a read-only first statement followed by a side-effecting second statement, such as `SELECT 1; RENAME TABLE ...`. The validator accepts the payload because it starts with `SELECT` and because side-effecting MySQL statements such as `RENAME TABLE`, `SET`, `RESET`, `LOCK TABLES`, and `KILL` are not rejected by the blacklist. In a live MCP runtime test, the `/mcp/query` endpoint accepted a `run_sql_readonly` request. The MCP response reported success for the first `SELECT`, and direct backend verification showed that the table had actually been renamed. This violates the endpoint's read-only security contract and lets an MCP caller perform backend writes or administrative SQL, limited by the configured MCP target account's database privileges. Version 3.0.9 contains a fix. Other operator mitigations include: keeping MCP disabled unless required; setting a non-empty `mcp-query_endpoint_auth` token before exposing `/mcp/query`; restricting MCP listener network exposure; configuring MCP backend target credentials as database-level read-only users; and adding temporary MCP query rules to block obvious multi-statement patterns.

Referencias oficiales

CVEs relacionados

CVECVSSEPSSKEVRExplotTítuloVis.

CVE-2026-850479.6 CRÍ

—

———Improper input validation in Transactions Platform in Google Chrome on on iOS prior to 152.0.7977.82 allowed a remote attacker to potentially execute arbitrary code outside the sandbox via a crafted HTML page. (Chromium security severity: Medium)3h CVE-2026-85239—

—

———A vulnerability in MISP's event template handling allowed an authenticated user with permission to create or modify event templates to bypass validation of the template definition field. The EventTemplate::beforeValidate() method only performed semantic validation when the supplied definition was already represented as an array. If a caller instead supplied a pre-encoded string, including malformed JSON or JSON representing an unexpected data type, the value bypassed validateDefinition() and only needed to satisfy the generic notBlank validation rule. As a result, an invalid event template definition could be stored persistently in the database. When event templates were subsequently retrieved, EventTemplate::afterFind() attempted to decode the stored definition using JsonTool::decode() without handling decoding failures. A definition containing invalid JSON could therefore trigger an exception during retrieval. Because the event template index is available to all authenticated users, a single malicious or malformed template could make the event template listing and other functionality relying on EventTemplate queries return HTTP 500 errors until the offending database row was manually repaired. Valid JSON representing an unexpected type, rather than the expected JSON object, could similarly result in invalid data reaching downstream consumers. The vulnerability can therefore be exploited by a user capable of saving event templates to persist malformed template data and cause a persistent denial of service against event-template functionality for other users. The patch enforces that event template definitions must be supplied as structured objects before saving and always applies semantic validation. On retrieval, malformed JSON and definitions that do not decode to the expected structure are caught, logged, and replaced with an empty definition, preventing a malformed database entry from breaking all event template queries. Poisoning doesn't seem reachable according to the lead developer.15h CVE-2026-85230—

—

———A persistent unsafe URL injection vulnerability exists in the MISP dashboard ButtonWidget configuration. Dashboard widget URLs were validated only when the widget was rendered and were not validated when the configuration was saved. As a result, an authenticated user able to modify dashboard widget settings could persist arbitrary URL values, including URLs using the javascript: scheme, through either of the dashboard settings persistence paths. A malicious javascript: URL stored in a dashboard button could potentially result in client-side script execution in the MISP security context if the value reached a rendering or navigation path without the existing runtime validation. Such execution could allow an attacker to perform actions with the privileges of the affected user or access information available to their MISP session. The practical exploitability of this issue is reduced by the fact that MISP already applied URL validation at render time, which neutralized known malicious values before they were presented to the user. The vulnerability therefore represents a persistence-layer validation gap and a defense-in-depth weakness rather than evidence of a direct bypass of the existing rendering protection. The patch introduces a canonical url schema type and validates dashboard widget configuration before it is persisted through either settings save mechanism. ButtonWidget URLs must now be strings resolving to an absolute path on the current MISP instance or a full URL with the same origin. Values using javascript:, external origins, malformed URL forms, and non-string values are rejected at save time.15h CVE-2026-85170—

—

———n8n versions before 1.123.73, 2.35.4, and 2.36.2 pass message content in the Gmail (v1) and Brevo nodes to the mail composer without verifying it is a string. An authenticated user able to run a workflow can supply an expression that resolves to an object carrying a path or href property, causing the composer to read a local file accessible to the n8n process or fetch an internal URL (SSRF) and attach the result to the outgoing message.18h CVE-2026-498304.4 MED

—

——0DSpace open source software is a repository application which provides durable access to digital resources. Prior to versions 7.6.7, 8.4, 9.3, and 10.0, when ingesting an aggregated ORE resource by URI (using the OAI-ORE Harvester), the ORE Ingestion Crosswalk does not validate the URI scheme. This may allow for local file inclusion via malicious paths like file:///etc/passwd. The attacker MUST already have DSpace collection administrator privileges in order to perform the attack. This issue has been patched in versions 7.6.7, 8.4, 9.3, and 10.0.16h CVE-2026-53600—

—

——0async-tar is a tar archive reading/writing library for async Rust. Prior to version 0.6.1, async-tar mis-applies a buffered PAX size extension to an intermediary extension header (a GNU longname L, a GNU longlink K, or a PAX x/g header) instead of to the next file entry. POSIX requires a PAX extended-header record set to describe the next file entry, never an intervening extension header. Because poll_next_raw (src/archive.rs) threads the buffered PAX records into the size computation of whatever raw header it reads next — and that header can be an intermediary L — the stream cursor is advanced by an attacker-chosen amount when the L body is consumed. The parser then desyncs relative to a POSIX-correct tar parser (e.g. GNU tar), reading subsequent bytes at the wrong block boundary. This issue has been patched in version 0.6.1.2d