CVE Vulnerability Catalog

In the guzzlehttp/psr7 library prior to version 2.12.3, the Uri::assertValidHost() function does not reject invalid URI host components that may contain authority delimiters, embedded ports, or malformed IPv6 brackets. This causes Uri::getHost() to potentially return a value different from the actual URI authority, which can lead to errors in security or routing decisions.

The vulnerability allows de-anonymization of TLS handshakes using Encrypted Client Hello (ECH). A passive network observer can read pre-shared key (PSK) identities in the unencrypted part of the client hello.

In U-Boot through 2026.04-rc3, an out-of-bounds read vulnerability exists in tcp_rx_state_machine() (net/tcp.c) when CONFIG_PROT_TCP is enabled. A remote attacker can craft a malicious packet with mismatched IP total length and TCP data offset field, causing reads beyond TCP segment boundaries and potentially disrupting connection state.

A vulnerability in the protobufjs library allows infinite looping when parsing specially crafted .proto files. The lack of end-of-input checking when processing options causes parse, Root.load, or Root.loadSync to loop indefinitely.

A vulnerability in the protobufjs library (versions 8.2.0 to 8.6.5) allows prototype pollution via parsing map entries in Text Format. An attacker can inject a map key named __proto__, which modifies the object prototype instead of creating a map entry.

The node-tar library before version 7.5.17 does not strip NUL bytes from PAX path and linkpath records, allowing a crafted archive to cause an uncaught exception and terminate the Node.js process.

The node-tar library before version 7.5.18 coerces all-digit PAX path and linkpath values to JavaScript numbers, causing a TypeError during path processing. The issue is fixed in version 7.5.18.

A vulnerability in the js-yaml library (versions 5.0.0 through 5.2.0) causes quadratic CPU time when parsing YAML documents with merge keys enabled. A chain of mappings where each entry merges the previous one leads to excessive CPU load.

AFFiNE's histories GraphQL field fails to validate Doc.Read permission before exposing document edit history, allowing authenticated workspace members to retrieve restricted content timelines. Attackers can supply arbitrary document GUIDs to access full edit histories including user names, emails, and timestamps of private pages they lack access to.

CyberChef before version 11.2.0 contains a prototype pollution vulnerability in the Series Chart operation, which accepts the __proto__ key when parsing user-supplied CSV. This can be chained with other operations like Parse UDP to inject malicious JavaScript into HTML output.

A vulnerability in Portainer Community Edition allows an unauthenticated attacker to access restore and administrator initialization endpoints within five minutes of starting an uninitialized instance. The attacker can restore a crafted backup or create the first administrator account, gaining full administrative access.

A vulnerability in ToolJet before 3.20.180 allows shell command injection via pull request comments. The CI runner executes code from the comment, enabling theft of deployment secrets.

A flaw was found in the gorch service template, part of the trustyai-service-operator. Even when authentication is enabled, the gorch service exposes unproxied orchestrator and detector metrics ports, allowing any pod on the cluster network to directly access these ports, bypassing the kube-rbac-proxy and its authentication mechanisms.

A vulnerability in SeaweedFS versions 4.08 through 4.33 allows an authenticated low-privileged S3 user to enumerate administrator-owned table bucket names and ARNs. The issue stems from incorrect routing of SigV4 requests to the S3Tables management API, where authorization collapses account-less S3 identities into the shared admin account.

A vulnerability in File Browser before version 2.63.16 allows an authenticated user with Create and Modify permissions to create attacker-controlled files outside the user's scope. The issue arises because ScopedFs validates the nearest existing ancestor of a dangling symlink as in scope and then follows the symlink during file creation.

A flaw in the TrustyAI Service Operator allows unauthenticated access to communication channels when deploying services like gorch or NemoGuardrails with a specific security setting disabled. This enables other programs within the cluster to access AI guardrails and orchestrator without proper authorization.

A vulnerability was found in Harness up to version 2.28.2 in the getAuthorizedSpaces function of the file app/api/controller/gitspace/list_all.go within the gitspaces Endpoint component. Manipulation of this endpoint can lead to authorization bypass. The attack can be executed remotely and the exploit has been publicly disclosed.

In MISP, the importModule() function used getEnabledModule() to resolve import modules by name but did not enforce per-organization restrictions. An authenticated user from a restricted organization could still invoke the module directly if they knew its name.

An authorization bypass in MISP's EventsController::importModule() allowed authenticated users or read-only API keys with event view access to persist data to events they were not allowed to modify. The write path did not verify event modification rights when processing misp_standard format results.

A stored XSS vulnerability in AVideo (Meet plugin) allows an attacker to inject malicious JavaScript via the User-Agent header when joining a public meeting. The payload is stored and later executed in the admin or host's browser session when viewing the participant list.