CVE Database

OGX (formerly Llama Stack, affected at commit fbe8e0f) contains an unauthenticated server-side request forgery vulnerability in the OpenAI-compatible POST /v1/responses endpoint. MCP tool definitions accept a server_url parameter (along with headers and authorization values) that is fetched server-side without destination validation; the existing validate_url_not_private() guard used for other URL inputs is not applied to server_url. On the default starter configuration, which runs without authentication, a remote unauthenticated attacker can cause the server to open connections to arbitrary internal addresses (including cloud metadata endpoints such as http://169.254.169.254/) and forward attacker-supplied headers and bearer tokens to those destinations.

Chroma 1.5.9 fails to validate maximum bounds on HNSW index parameters max_neighbors, ef_construction, and ef_search in collection-create requests. Unauthenticated attackers can supply arbitrarily large parameter values to exhaust server memory and cause denial of service during index compaction.

cli-mcp-server 0.2.5 contains a command allowlist bypass vulnerability in the _validate_command_with_operators function when ALLOW_SHELL_OPERATORS is enabled. Attackers can use shell command substitution syntax like $(...) or backticks to execute non-allowlisted commands that bypass the ALLOWED_COMMANDS validation check.

Trigger.dev versions before 4.5.2 fail to validate environment membership during run replay operations, allowing authenticated attackers to inject task runs into arbitrary environments. Attackers can replay their own runs into other organizations' or projects' environments to consume victim resources and pollute run history.

git-mcp-server 2.15.1 contains an argument injection vulnerability in the ref and object parameters of git_log, git_diff, and git_show tools that lack leading-dash validation. Attackers can inject git command-line options like --output= to write files outside the repository to arbitrary paths accessible by the process.

sift (sift.js) 17.1.3 enumerates query keys with for...in, which walks the object prototype chain, and dispatches any matched operator key including $where. The $where operation compiles a string value into a function using new Function unless CSP_ENABLED is set (not set by default). As a result, if a prototype-pollution primitive elsewhere in the process sets Object.prototype.$where to a malicious string, even benign filter calls such as sift({}) execute arbitrary JavaScript. Additionally, passing an untrusted query object containing a string $where directly to sift results in code execution under the default configuration.

goose 1.37.0 executes arbitrary commands from recipe stdio extensions and retry.checks without security inspection. Attackers can distribute malicious recipes that execute shell commands as the user running goose, bypassing the recipe security scan which does not inspect extensions or retry configurations.

Postgres MCP Pro 0.3.0 contains a restricted-mode bypass vulnerability where function-name validation is not applied to RangeFunction nodes in FROM clauses. Attackers can execute file-reading functions like pg_read_file through FROM-clause syntax to read arbitrary files despite restricted-mode protections.

AppFlowy-Cloud 0.9.64 fails to verify that requested collab objects belong to the workspace in authorization checks, allowing attackers to access documents and database rows across workspaces. Attackers can supply a victim's object ID with their own workspace ID to bypass access controls and read, modify, or delete cross-workspace data.

Douyin_TikTok_Download_API through 4.1.2 contains a server-side request forgery vulnerability in the /api/download and /api/hybrid/video_data endpoints that allows unauthenticated attackers to fetch arbitrary URLs by supplying a url query parameter. Attackers can request internal services including cloud metadata endpoints and retrieve response bodies containing sensitive credentials through error messages.

Blinko 1.8.7 contains an authorization bypass (IDOR) vulnerability in multiple tRPC procedures (message.list, message.update, message.delete, message.clearAfter in server/routerTrpc/message.ts and conversation.clearMessages in server/routerTrpc/conversation.ts). Although these procedures require authentication, they query the database by caller-supplied conversation or message ID without verifying that the resource belongs to the requesting account. Any authenticated user can therefore read another user's full AI chat history, modify individual message content, and delete or wipe entire conversations by enumerating sequential integer IDs.

firecrawl-mcp-server 3.20.2 contains an arbitrary local file read vulnerability in the firecrawl_parse tool that accepts unconstrained filePath arguments without directory containment validation. Attackers can supply absolute paths or directory traversal sequences to read sensitive files like credentials and environment variables, which are then uploaded and returned to the model context.

Allocation of Resources Without Limits or Throttling vulnerability in elixir-mint mint allows a remote HTTP server to exhaust memory on the client host and cause a denial of service. Two HTTP/1 response-parser states accumulate server data without any cap. In lib/mint/http1.ex, decode_status_line/4 stores the unconsumed data in conn.buffer when the status line is incomplete, and decode_body/5 does the same for an unterminated chunk-extension line. Both wait for a CRLF the server never has to send, and conn.buffer is prepended to every subsequent socket message. The :max_header_list_size budget is wired only into decode_headers/5 and decode_trailer_headers/4, so neither of these states is covered by it. A malicious server, or one reached through an attacker-controlled redirect or a fetched URL, streams bytes indefinitely until the BEAM node is killed by the operating system out-of-memory handler. The chunk-extension variant is reached after a valid status line and a complete, valid header section, so an intermediary inspecting only headers sees an ordinary 200 response. This issue affects mint: from 0.1.0 before 1.10.0.

IBM App Connect Enterprise 13.0.1.0 through 13.0.8.1, and 12.0.1.0 through 12.0.12.28 and IBM Integration Bus for z/OS 10.1.0.0 through 10.1.0.7 SAP Adapter is vulnerable to an XML external entity (XXE) attack.

Acunetix 25.11.251107123 for Windows contains a local privilege escalation vulnerability in the Web Vulnerability Scanning Engine (wvsc.exe) that allows low-privileged local attackers to execute arbitrary code as SYSTEM by exploiting a missing hardcoded directory path for OpenSSL-related files. Attackers can create the missing directory, place a malicious file at the expected path, and cause the SYSTEM-level wvsc.exe process to load and execute it, resulting in full privilege escalation.

Observable response discrepancy vulnerability in GastroMenum GastroMenum Web Panel allows Account Footprinting. This issue affects GastroMenum Web Panel: before 31.08.2026.

** UNSUPPORTED WHEN ASSIGNED ** Improper Neutralization of Special Elements used in an SQL Command ('SQL Injection') vulnerability in Apache Griffin Hive Metastore Module.  This issue affects Apache Griffin Hive Metastore Module: all versions. As this project is retired, we do not plan to release a version that fixes this issue. Users are recommended to find an alternative or restrict access to the instance to trusted users. NOTE: This vulnerability only affects products that are no longer supported by the maintainer.

The LearnDash LMS plugin for WordPress is vulnerable to Unrestricted File Type Upload in versions up to and including 5.1.5. This is due to insufficient input validation in the 'learndash_fileupload_process' function, which iterates through an entire array and validates only the first file. This makes it possible for authenticated attackers, with subscriber-level access and above who are enrolled in a course with assignment uploads enabled, to upload arbitrary disallowed files, including PHP files, to the server's wp-content/uploads/learndash/assignments/ directory. The uploaded files can only be used for Remote Code Execution if default server configurations have been changed to allow for execution.

(Holloway) Chew, Kean Ho's Actualizer v1.2.0 and earlier contains a fail-open password validation vulnerability in the Alpha user and root user password loops of Shell/debian-minbase-install.sh. The installer invokes mkpasswd to generate yescrypt password hashes but does not check the command's return value and unconditionally accepts the result. If mkpasswd fails to generate a yescrypt hash, for example because an incompatible mkpasswd implementation or an environment without yescrypt support is used, the resulting password hash variable can be empty and the build proceeds. The resulting image can therefore contain empty password fields for the root and alpha accounts, potentially permitting passwordless authentication depending on the authentication configuration.

A vulnerability was detected in code-projects Vehicle Management System 1.0. The affected element is an unknown function of the file /busprofile.php. Performing a manipulation of the argument busid results in sql injection. It is possible to initiate the attack remotely. The exploit is now public and may be used.

Improper neutralization of special elements used in an SQL command ('SQL injection') vulnerability in TAC Information Services Internal and External Trade Inc. GOLDENHORN ONEIT allows Blind SQL Injection. This issue affects GOLDENHORN ONEIT: before Göbeklitepe.

snipe-it versions before 8.6.3 contain an authorization bypass vulnerability in the bulk delete functionality that allows restricted users to soft-delete users outside their authorized scope. Attackers can include unauthorized user IDs in bulk delete requests to bypass instance-level restrictions and modify or disable accounts they should not access.

Snipe-IT versions before 8.6.2 contain an authorization bypass vulnerability in checkout-acceptance report actions when Full Multiple Company Support is enabled. Authenticated users with reports.view permission can enumerate sequential acceptance IDs and soft-delete or trigger reminder emails for acceptances belonging to other companies by exploiting a null check on the legacy users.company_id column.

OpenPanel before 2.3.0 contains an unauthenticated server-side request forgery vulnerability in the GET /tools/site-checker endpoint that accepts a fully client-controlled URL parameter with no private IP filtering or DNS-rebinding protection. Attackers can make the OpenPanel server issue requests to internal services, localhost, and cloud metadata endpoints, reading internal HTTP response titles, headers, status codes, and SSL certificate information.

OpenPanel before 2.3.0 contains a cross-site scripting vulnerability in the unauthenticated favicon proxy endpoint GET /misc/favicon that allows remote attackers to execute scripts by supplying an SVG file URL. Attackers can host malicious SVG files with embedded scripts that execute in the victim's browser on the API origin, enabling same-origin credentialed requests to authenticated endpoints.

OpenPanel before 2.3.0 contains an unauthenticated server-side request forgery vulnerability in the /misc/favicon and /misc/og endpoints that accept an attacker-supplied url parameter with insufficient validation. Attackers can force the API to fetch arbitrary internal hosts and cloud metadata endpoints, with small responses returned verbatim enabling credential theft and internal service enumeration.

OpenPanel before 2.3.0 fails to properly validate chart formula expressions, allowing authenticated project members with read access to execute arbitrary code by recovering the native JavaScript Function constructor through mathjs matrix objects. Attackers can use the recovered constructor to load Node.js built-ins and execute operating system commands with the privileges of the API process, bypassing organization authorization boundaries.

Openpanel before 2.3.0 contains an unauthenticated full-read server-side request forgery (SSRF) vulnerability in the GET /tools/site-checker endpoint (apps/api/src/controllers/tools.controller.ts). The endpoint passes a user-supplied url query parameter to fetchWithRedirects() and performs server-side HTTP requests to arbitrary URLs without any SSRF/IP validation. An unauthenticated remote attacker can access cloud instance metadata endpoints, probe internal services, scan internal network ports, and read returned content (status code, page size, timing, and parsed HTML metadata), and leak internal IP addresses (via getIPInfo() to a third party).

Grav before 2.0.18 (affected versions <= 2.0.17) contains a remote code execution vulnerability in the Twig sort filter. The sortFunc wrapper in GravExtension.php hardcodes Twig's isSandboxed argument to false, so unlike |map/|filter/|reduce, |sort accepts a plain function name inside the sandbox; the remaining denylist misses spl_autoload, which performs a PHP include. An authenticated user with only page-write rights (admin.pages or api.pages.write) can supply a crafted payload (e.g., via form frontmatter rendered by the Email plugin) that invokes spl_autoload through the sort filter, resulting in arbitrary PHP execution as the web server user.

Grav Shortcode Core before 6.2.5 contains stored cross-site scripting vulnerabilities in the [lorem] tag parameter and [details] summary parameter that are written to rendered pages without escaping. Attackers with page-edit access can inject arbitrary HTML and JavaScript that executes in the browsers of all page visitors, including administrators.