ArcadeDB versions from 21.10.1 up to but not including 26.11.1 contain a resource-limit bypass vulnerability. This flaw allows restricted users to evade per-group readTimeout limits by exploiting the way cached SQL statements and plans are shared. Attackers can execute SQL text cached by unrestricted users to bypass timeout restrictions or impose their own timeout and limit on privileged users such as root.
ArcadeDB versions prior to 26.11.1 have a denial of service vulnerability in the Redis protocol handler. The flaw occurs because the handler neither replies to nor closes a pre-authentication connection when it receives an array whose first element is not a string. This allows unauthenticated attackers to keep connections open indefinitely by repeatedly sending such arrays just within the 30-second socket idle timeout, exhausting RedisNetworkExecutor threads.
ArcadeDB versions from 26.2.1 up to but not including 26.11.1 have a vulnerability where the MCPToolUtils.resolveDatabase method reveals whether a database exists before authorization, returning different errors for 'not authorized' versus 'does not exist'. Additionally, a failed Bolt LOGON command leaves the previous principal bound after a RESET. This allows authenticated MCP users to enumerate database names they are not authorized to access by probing with certain tools.
ArcadeDB versions from 26.11.1-SNAPSHOT up to but not including 26.11.1 contain an authorization bypass vulnerability. This flaw allows authenticated users to bypass per-type access control checks when opening a database on demand. Exploitation involves sending a specially crafted POST request to a closed database, enabling unauthorized read and write access to data types.
ArcadeDB versions before 26.11.1 have an authorization bypass vulnerability that allows authenticated users to read schema definitions of types for which they lack access permissions. This bypass occurs through listings that ignore per-type access control lists (ACLs), enabling attackers to query the PostgreSQL wire catalog, GraphQL introspection, or MCP get_schema to obtain sensitive schema details such as property names, indexes, inheritance, and materialized view SQL.
actix-http versions from 3.18.10 up to but not including 3.20.2 contain an HTTP request smuggling vulnerability in the chunked decoder. The flaw arises because the decoder treats a bare line feed (LF) after the last chunk as the start of a trailer section. This allows attackers to send a specially crafted body ending with 0\r\n\n through a connection-reusing proxy, causing the next request's body to be misinterpreted as new requests. This can bypass web application firewalls (WAF) or other upstream protections.
actix-files versions prior to 0.7.4 have an information disclosure vulnerability that allows remote unauthenticated attackers to read hidden files by exploiting Windows 8.3 short-name aliases. This bypasses the leading-dot check in actix_files::Files, enabling access to sensitive files such as .env and .git contents.
Rclone versions from 1.49.0 up to but not including 1.76.0 have an authentication vulnerability in the Proxy.Call function. This flaw allows users whose access has been revoked to retain access by reconnecting within five minutes using cached authentication responses. The issue affects serve modes for sftp, ftp, http, and webdav protocols.
ArcadeDB native-image server builds from 26.8.1 up to but not including 26.11.1 have an authorization bypass vulnerability. This flaw allows authenticated clients to perform write operations through query paths intended to be read-only by nesting statements within SELECT queries. The vulnerability arises due to missing reflection metadata causing the SQL AST inspector to see an empty abstract syntax tree, enabling INSERT statements in projections or LET clauses to persist via various query endpoints and modes.
Rclone versions from 1.46.0 up to but not including 1.76.0 have a memory exhaustion vulnerability in the remote control (RC) server. This flaw allows unauthenticated clients to crash the rclone process by sending oversized JSON payloads to protected RC methods. The vulnerability arises because the server decodes the JSON body before performing authorization checks, leading to out-of-memory termination and interruption of ongoing transfers.
Rclone versions from 1.56.0 up to but not including 1.76.0 have a timing side-channel vulnerability in the htpasswd authentication middleware. This flaw allows unauthenticated attackers to enumerate valid usernames by sending HTTP Basic Auth requests with incorrect passwords and measuring response times. The vulnerability arises because bcrypt verification is only performed for existing users, causing measurable timing differences. The vulnerability has a medium severity with a CVSS score of 5.3.
A denial of service vulnerability exists in actix-http versions before 3.20.3. Unauthenticated remote attackers can exploit this by sending large-window Brotli encoded request bodies, causing excessive memory allocation and potentially crashing the server.
ArcadeDB versions from 26.7.1 up to but not including 26.11.1 have an information disclosure vulnerability in the LSMTreeFullTextIndex component. This flaw allows authenticated users with read access to a supertype, but not its subtype, to infer the existence of hidden subtype documents by running EXPLAIN or PROFILE on full-text queries. The vulnerability arises because BM25 corpus statistics are computed over all buckets of the indexed type without enforcing bucket access control on subtypes.
ArcadeDB versions from 25.8.1 up to but not including 26.11.1 contain an incorrect authorization vulnerability related to per-type ACLs. The issue arises because ACLs are keyed by type name, and renaming types does not update the permission configuration. This allows authenticated users with only updateSchema privileges to rename a denied type and gain unauthorized read, delete, and truncate access. The unauthorized access is immediate for TIMESERIES types and delayed until the next security rebuild for record-backed types.
ArcadeDB versions from 26.3.1 up to but not including 26.11.1 contain a vulnerability that allows server administrators to bypass read-only restrictions in the MCP read-only query tool. This is achieved by submitting Groovy-fallback Gremlin scripts that perform write operations, which are incorrectly classified as read-only and thus allowed to persist. The vulnerability affects the GremlinQueryEngine.analyze() method, which fails to block these write operations in a read-only context.
better-auth versions 1.3.0 before 1.7.16 and 1.8.0-beta.0 before 1.8.0-beta.5 have an authorization vulnerability in the admin plugin's /admin/update-user endpoint. This flaw allows users with the user:update role to modify another user's phoneNumber and phoneNumberVerified fields. Exploiting this, an attacker can assign their own verified phone number to a higher-privileged account and gain access via phone OTP authentication.
better-auth versions 1.4.0 before 1.7.16 and 1.8.0-beta.0 before 1.8.0-beta.5 have an improper authentication vulnerability. This flaw allows attackers to reuse email-change links issued for their own accounts to hijack victim accounts by changing the victim's email to one controlled by the attacker.
A vulnerability in better-auth and @better-auth/passkey allows revoked sessions to remain valid due to non-atomic updates of secondary storage session lists and reliance on cookie-cached sessions. This flaw enables an attacker with a valid session to bypass bulk revocation or bans, or to use a revoked cached session to register a passkey or create a persistent session.
SurrealDB versions from 1.0.0 up to but not including 4.0.0 contain an authorization bypass vulnerability. Authenticated users can exploit this flaw by binding protected variables ($access, $auth, and $token) as FOR loop variables in queries. This allows attackers to bypass table permission checks and perform unauthorized read, modify, or delete operations with VIEWER privileges.
Better Auth versions before 1.7.16 and 1.8.0-beta.5 contain an OAuth sign-in authentication flaw. The flaw causes unverified provider emails, including those with the string "false" in the email_verified claim, to be treated as verified. Additionally, generic OAuth profile and ID-token subject mismatches are unchecked, and Zoom provider options are ignored. This allows attackers who control a victim's unverified email at an affected provider to sign in, link to the victim's local account, and obtain the victim's session.