CVE Database

ArcadeDB is a Multi-Model DBMS. Prior to 26.7.1, a reader-role user can submit POST /api/v1/command/{database} with language: js because PolyglotQueryEngine.command, PolyglotQueryEngine.analyze, and PolyglotQueryEngine.registerFunctions do not enforce database-administrator authorization. GraalPolyglotEngine also permits scripts to bypass the allowedPackages whitelist by reflecting from the bound database object through database.getClass().getClassLoader().loadClass to arbitrary host classes. These cooperating defects allow a read-only database user to read arbitrary host files outside the database scope. Process creation is already blocked, so OS command execution is not confirmed. The issue is distinct from CVE-2026-44221, CVE-2026-54076, and CVE-2026-54077. This issue is fixed in version 26.7.1.

FrontMCP is a TypeScript-first framework for the Model Context Protocol (MCP). From mcp-from-openapi 2.3.0 until 2.5.0 and from frontmcp and @frontmcp/adapters 1.2.1 until 1.5.0, libs/adapters/src/openapi/openapi.adapter.ts loadOpenAPISpec() forwards untrusted OpenAPI url and spec inputs and loadOptions.refResolution to OpenAPIToolGenerator.fromURL() and OpenAPIToolGenerator.fromJSON(). The external $ref guard checks parsed hostname strings without resolving addresses, pinning validated addresses, revalidating redirect targets, or normalizing IPv4-mapped IPv6. An authenticated user who can import or configure an OpenAPI specification in a hosted or multi-user deployment can use DNS-to-loopback resolution, redirect-to-loopback behavior, or IPv4-mapped IPv6 loopback forms to cause backend-origin requests to internal services. This can expose internal administrative APIs, metadata-like services, and other private network endpoints. The practical impact is lower when only a trusted local administrator can configure OpenAPI specs, and disabling external reference protocols prevents the external $ref request. This issue is fixed in mcp-from-openapi 2.5.0 and frontmcp and @frontmcp/adapters 1.5.0.

Payload Plugins is a collection of plugins designed to enhance Payload CMS. In 0.7.0, @jhb.software/payload-alt-text-plugin exposes POST /api/alt-text-plugin/generate and POST /api/alt-text-plugin/bulk with a default guard that accepts any authenticated user, while alt-text/src/endpoints/generateAltText.ts and alt-text/src/endpoints/bulkGenerateAltTexts.ts call req.payload.findByID and req.payload.update without overrideAccess: false. Payload therefore defaults overrideAccess to true and skips the target collection's read and update access functions. An authenticated low-privilege user can supply id, collection, locale, and update values to read arbitrary protected upload documents and overwrite their alt and keywords fields, even when the collection permits those operations only to administrators. A control Local API call with overrideAccess: false is denied, confirming that the plugin endpoint bypasses otherwise effective collection rules. This vulnerability is fixed in 0.8.0.

webhookd is a minimalist webhook server that triggers shell scripts and external processes through HTTP requests. Prior to 1.22.0, webhookd deployments without htpasswd authentication forwarded all incoming HTTP headers through HTTPParamsToShellVars in pkg/api/index.go into the hook script environment without an allowlist. When an upstream reverse proxy did not strip a client-supplied X-WebAuthn-User header and a hook script trusted that variable for identity or privilege, a remote unauthenticated attacker could spoof another user, bypass script security controls, and access or modify resources available to the impersonated identity. This issue is fixed in version 1.22.0.

ToolHive is a utility designed to simplify the deployment and management of Model Context Protocol (MCP) servers. Prior to 0.31.0, remote.Handler.Authenticate in pkg/auth/remote/handler.go invokes discovery.DetectAuthenticationFromServer in pkg/auth/discovery/discovery.go, whose host-side HTTP clients trust remote-server-controlled authentication discovery destinations, follow redirects without host or scheme restrictions, and do not consistently block private addresses. A malicious or compromised remote MCP server can place a resource_metadata URL in WWW-Authenticate for ParseWWWAuthenticate to extract, after which FetchResourceMetadata, OIDC issuer discovery, and well-known discovery can issue GET requests to link-local, RFC1918, or other internal services outside the server's container. The user connects to a server that the user intends to use, but the server controls the internal destination; this path does not depend on the separate NAT64 guard omission because the affected clients did not invoke IsPrivateIP, and it differs from the DCR resolver path that already refused redirects. The demonstrated primitive reaches internal-only HTTP services and reachability or error oracles, and can retrieve AWS metadata credentials where IMDSv1 accepts an unauthenticated GET, while IMDSv2 and GCP metadata prerequisites are not satisfied by the demonstrated request. This vulnerability is fixed in 0.31.0.

MCP Gateway allows easy and secure running and deployment of MCP servers. From 0.21.0 until 0.42.2, Docker MCP Gateway YAML-unmarshalled the attacker-controlled io.docker.server.metadata OCI image label into the broad catalog.Server structure for direct docker:// references and catalog snapshot imports in pkg/oci/self_contained.go and pkg/workingset/workingset.go. Runtime-shaping fields including Volumes, User, and ExtraHosts were then appended to the docker run argument vector without an origin allowlist, allowing a malicious image author to request host filesystem or Docker socket mounts and UID 0 execution when a victim selected or pulled the image. This container-creation-time boundary bypass can execute arbitrary code on the host and is not prevented by no-new-privileges because no in-container privilege escalation is required. This issue is fixed in version 0.42.2.

GeoNetwork is a catalog application to manage spatially referenced resources. Prior to 4.2.17 and 4.4.12, POST /api/tools/ogc/sld accepted a caller-supplied WMS server URL and performed a server-side HTTP GET without destination validation. An anonymous attacker could make the GeoNetwork server send requests to internal hosts that are not publicly reachable. When the outbound response was XML, the endpoint could store and return the fetched body, making the request forgery non-blind and enabling internal data disclosure, authorization bypass, and network reconnaissance. This issue is fixed in versions 4.2.17 and 4.4.12.

qbee transport is a remote access transport protocol implementation. Prior to 1.26.25, the extractTar routine uses strictly lexical path validation that does not account for on-disk symlinks created earlier in the extraction process. A crafted tar archive can use a symlink chain to write or overwrite files one directory level above the intended extraction path. When qbee-agent performs the extraction with root privileges, this permits a root-privileged file write outside the intended destination. This issue is fixed in version 1.26.25.

OpenBao is an open source identity-based secrets management system. Prior to 2.5.5, an authenticated OpenBao caller with write access to transit/keys/* could terminate the server process by setting derived to true while the type parameter selected rsa-, ecdsa-, or ed25519. The Transit policy creation path in builtin/logical/transit/backend.go and sdk/helper/keysutil/policy.go could reach an error path that double-unlocked a mutex while handling this invalid asymmetric derived-key combination, causing a panic, no HTTP response, process exit, and denial of service. JSON and HCL key-creation requests can express the triggering combination. This issue is fixed in version 2.5.5.

OpenBao is an open source identity-based secrets management system. Prior to 2.5.5, OpenBao users granted capabilities on /sys/namespaces/root within a non-root namespace could exploit special handling of the literal root path in namespace canonicalization. The /sys/namespaces/* endpoint family can resolve its containing namespace through a path prefix or X-Vault-Namespace header. ACL checks occurred before root canonicalized to an empty path, causing /sys/namespaces/root to resolve to the system backend's containing namespace and allowing permitted lookups, deletion, locking, or custom metadata changes against that direct containing namespace. The root namespace and arbitrary unrelated namespaces are not affected, and available operations depend on the capabilities granted on the path and subpaths such as /api-lock. This issue is fixed in version 2.5.5.

OpenBao is an open source identity-based secrets management system. Prior to 2.5.5, an OpenBao user with access to sys/leases/revoke/:lease_id in one namespace could revoke a lease in another namespace when the foreign lease_id was known, bypassing namespace ACL isolation. The affected lease lookup routing in vault/expiration.go allowed FetchLeaseInfo and loadEntry to resolve cached or stored lease data outside the request namespace, allowing a tenant that intentionally disclosed a lease identifier to have the lease and its underlying credential revoked by another tenant. This issue is fixed in version 2.5.5.

OpenBao is an open source identity-based secrets management system. Prior to 2.5.5, OpenBao used EscapeLDAPValue, an RFC 4514 distinguished-name escaping function, where RFC 4515 LDAP search-filter escaping was required in sdk/helper/ldaputil/client.go GetUserDN. With the LDAP authentication backend configured for an Active Directory UPNDomain path or UserDN and UserAttr binding, an attacker-controlled username containing filter metacharacters could alter the search predicate and select a different directory entry because EscapeLDAPValue does not neutralize the characters handled by ldap.EscapeFilter. A resulting token could be associated with another LDAP identity and gain access to secrets, policies, or modification capabilities assigned to that identity. This issue is fixed in version 2.5.5.

The OpenTelemetry Collector Contrib repository contains components for the OpenTelemetry Collector. Prior to 0.151.0, the githubreceiver validates the receiver/githubreceiver/config.go RequiredHeaders configuration at startup, but receiver/githubreceiver/trace_receiver.go handleReq() does not check those headers on incoming webhook requests. An unauthenticated sender can therefore bypass an operator's required_headers authentication control and submit arbitrary webhook payloads. When the Secret field is empty, github.ValidatePayload also skips HMAC validation, leaving the webhook endpoint without either configured authentication mechanism. Successful exploitation can inject fabricated CI/CD trace data into the observability pipeline. This issue is fixed in version 0.151.0.

Capsule is a multi-tenancy and policy-based framework for Kubernetes. From 0.13.0 until 0.13.6, charts/capsule/templates/configuration.yaml configures the validating webhook with namespace/finalize instead of the Kubernetes resource name namespaces/finalize. A user with namespaces/finalize RBAC can send a PUT request to /api/v1/namespaces/{namespace}/finalize, and the singular rule never matches the plural resource, so the validating webhook is not invoked and the user can change the namespace tenant label. matchPolicy: Equivalent does not compensate because it handles API group and version equivalence rather than resource-name errors. This vulnerability is fixed in 0.13.6.

Kiwi TCMS is an open source test management system. Prior to 16.1, TestCase.extra_link and TestPlan.extra_link accepted unsanitized user input and rendered stored values verbatim, creating an opportunity for cross-site scripting. Official Docker images and unmodified Kiwi TCMS middleware send a Content-Security-Policy header that blocks inline JavaScript, making exploitation difficult in default deployments, while customized deployments that weaken those security settings may remain vulnerable. Version 16.1 properly sanitizes both fields and resets existing database records that do not validate to null. This issue is fixed in version 16.1.

Signal K Server is a server application that runs on a central hub in a boat. Prior to 2.28.0, makeRemoteRequest() in src/serverroutes.ts accepted attacker-controlled host, port, useTLS, and selfsignedcert parameters from the testSignalKConnection, requestAccess, and checkAccessRequest endpoints without validating the destination. When security was not configured, addAdminMiddleware() was a no-op in dummysecurity.ts, leaving all three endpoints accessible without authentication. The server could be forced to contact loopback, private, link-local, cloud metadata, or arbitrary external destinations, and selfsignedcert could disable certificate verification for outbound HTTPS requests. The checkAccessRequest endpoint also interpolated requestId into its destination path, allowing traversal to other paths on the selected host. Distinct success, connection-refused, and timeout responses enabled internal port scanning; returned response bodies enabled cloud metadata and internal-service data exfiltration; requestAccess enabled server-side POST requests with attacker-controlled JSON; and access to cluster-internal services could support lateral movement. This issue is fixed in version 2.28.0.

Surfio is a library for reading and writing surface files. Prior to 0.0.19, surfio does not correctly validate size fields in IRAP files, leading to a buffer overflow when untrusted files are parsed. The severity assumes surfio is used to parse untrusted files in a networking context such as a web service. This issue is fixed in version 0.0.19.

Kiwi TCMS is an open source test management system. Prior to 16.1, the account confirmation endpoint accepted an unvalidated next parameter, allowing an unauthenticated attacker to create a URL on a trusted Kiwi TCMS hostname that redirects a victim to an arbitrary external domain. The trusted origin can support credential-harvesting pages, bypass email security filters and link-reputation checks that allowlist the organization's domain, or deliver malware through a convincing account-confirmation lure. This issue is fixed in version 16.1.

ToolHive is a utility designed to simplify the deployment and management of Model Context Protocol (MCP) servers. Prior to 0.29.1, networking.IsPrivateIP in pkg/networking/utilities.go omits the IPv6 NAT64 prefixes 64:ff9b::/96 and 64:ff9b:1::/48, so NAT64 addresses embedding private, loopback, or link-local IPv4 targets are classified as public and allowed. The most direct attacker-controlled path begins when an external OAuth client supplies a client_id URL that CIMDStorageDecorator.GetClient routes through FetchClientMetadataDocument in pkg/oauthproto/cimd/fetch.go; protectedDialerControl in pkg/networking/http_client.go and validateHost in pkg/skills/gitresolver/reference.go share the defective classification but use operator-controlled or user-controlled destinations. On a ToolHive host behind a NAT64/DNS64 gateway, the gateway translates an allowed address such as 64:ff9b:1::a9fe:a9fe to 169.254.169.254, permitting blind probing of internal TCP or TLS reachability. The attacker-controlled CIMD path requires HTTPS, verifies certificates, and does not reflect response bodies, so the established impact is an internal reachability oracle rather than metadata credential exfiltration; the webhook client is not affected because it does not use this IP guard. This issue is fixed in version 0.29.1.

ArcadeDB is a Multi-Model DBMS. Prior to 26.6.1, the IMPORT DATABASE statement in engine/src/main/java/com/arcadedb/query/sql/parser/ImportDatabaseStatement.java did not require administrative privileges and passed its source to integration/src/main/java/com/arcadedb/integration/importer/SourceDiscovery.java without validation. An authenticated user with SQL command access through /api/v1/command or /api/v1/query can supply HTTP or HTTPS destinations to make server-side requests to internal services, or file:// paths to read files accessible to the server process and ingest the results as queryable records. The XML importer also permits DTD processing and external entities, enabling entity expansion. The root-only /api/v1/server administration endpoint is not affected. The fix requires updateSecurity permission, blocks local-network import destinations by default through arcadedb.server.security.importBlockLocalNetworks, supports the arcadedb.server.security.importAllowedLocalPaths file allow-list, and disables XML DTD processing and external entities. This issue is fixed in version 26.6.1.

ArcadeDB is a Multi-Model DBMS. Prior to 26.6.1, the fix for CVE-2026-44221 added an UPDATE_SCHEMA authorization check only to LocalDocumentType.createProperty, while the remaining public schema mutators in engine/src/main/java/com/arcadedb/schema/LocalDocumentType.java and engine/src/main/java/com/arcadedb/schema/LocalProperty.java remained unchecked. An authenticated identity, including a read-only API token without UPDATE_SCHEMA permission, can use DROP PROPERTY, ALTER TYPE, or ALTER PROPERTY through the database command/query HTTP endpoints to rename types, change inheritance, alter aliases or buckets, drop properties, and change property constraints. The issue does not directly disclose or write record data, but unauthorized schema mutation can corrupt the meaning of stored records and breach the documented permission model. This issue is fixed in version 26.6.1.

GitHacker is a tool that restores Git repositories from exposed .git directories. In 1.1.7 and earlier, add_head_file_tasks parses an attacker-controlled ref path from .git/HEAD and joins unvalidated path segments onto temp_dst/.git/logs/, allowing a malicious server to make GitHacker read an arbitrary local file when a victim runs the tool against the server's URL. add_hashes_parsed then scans the file for 40-character hexadecimal substrings and requests each match through .git/objects using the first two characters and remaining characters as path components, creating an attacker-observable existence oracle and disclosing matching hexadecimal fragments. Complete file contents are not returned, and the add_folder and add_task write path in shipped version 1.1.7 does not permit an escape. This issue is fixed in repository version 1.1.9.

SingularityCE and SingularityPRO are open source container platforms. Prior to SingularityCE 4.4.2 and SingularityPRO 4.3.9 and 4.1.14, incorrect path-string matching in the singularity.conf limit container paths directive allows a container in a sibling directory such as /data/safe-but-unsafe to be run when /data/safe is allowed under setuid mode. This permits a user to run a container from outside the administrator's configured path allowlist. Installations that do not use limit container paths are not affected. This issue is fixed in SingularityCE 4.4.2 and SingularityPRO 4.3.9 and 4.1.14.

Decidim is a participatory democracy framework. Prior to 0.32.0, a low-privilege process-scoped administrator or election editor with question-management rights can store HTML or script-bearing content in question.body. The question_title helper returns the translatable question body through html_safe without a sanitization boundary, causing stored script execution when visitors open public election pages or voting booth screens. The persisted script executes in visitors' browsers. The vulnerability is fixed in 0.32.0.

fluent-plugin-opentelemetry is a Fluentd input and output plugin for forwarding OpenTelemetry Protocol data. Prior to 0.5.3, the in_opentelemetry HTTP input read the entire incoming request body and decompressed payloads into memory without enforcing maximum size thresholds. When an OpenTelemetry ingestion endpoint was exposed to an untrusted network, an attacker could send an excessively large request or a highly compressed payload that expanded in memory. The resulting memory exhaustion could cause the operating system to terminate the Fluentd process, disrupting all log collection and forwarding on the affected node. This issue is fixed in version 0.5.3.

TOTOLINK X5000R V9.1.0cu.2415_B20250515 was discovered to contain a hardcoded password for root access.

Bucket Squatting in Google Cloud Gemini Enterprise Agent Platform SDK for Python versions prior to 1.166.1 allows an attacker to achieve Remote Code Execution (RCE) and tenant-project token theft.

Tornado before 6.4.1 contains a CRLF injection vulnerability in CurlAsyncHTTPClient that fails to reject carriage return and line feed characters in request headers. Attackers can inject CRLF sequences into header values to inject arbitrary headers or construct entirely new HTTP requests.

Tornado before 6.4.1 ignores duplicate Transfer-Encoding: chunked headers, treating requests as having no message body and parsing the chunked body as a subsequent request. Attackers can exploit this inconsistency when Tornado is deployed behind proxies to perform HTTP request smuggling, enabling access control bypass, cache poisoning, or connection desynchronization.

Tornado before 6.3.3 contains an HTTP request smuggling vulnerability due to improper parsing of Content-Length headers accepting non-standard characters. Attackers can send crafted HTTP requests with these characters to bypass proxy validation and smuggle requests when deployed behind certain proxies.