---
_-= Per source details. Do not edit below this line.=-_
## Source: amazon-inspector (d3c551a6031e48a54b550a604e2198a4491cf935407744bba8ef0c11c54086c4)
The package advertises EVM address validation but performs no local validation. Its single exported function validated(e) in index.js issues axios.get against a hardcoded destination and appends the caller-supplied argument to the URL, causing whatever the consuming application passes in (EVM addresses, keys, or arbitrary user data) to be transmitted to a remote host the caller never configured. The destination is deliberately hidden: a scrambled string array is joined, ROT13-transformed, and reversed at runtime to reconstruct https://api-ivalidte.vercel.app/app/api/ — a typosquat-shaped Vercel host that does not correspond to any documented service for this package. The obfuscation of the destination combined with the mismatch between the advertised purpose (local validation) and the actual behavior (remote relay to an author-controlled endpoint) is the silent-relay shape.
A vulnerability exists in the ktls_ocf_tls_cbc_decrypt() function when processing TLS 1.2 CBC records. The function incorrectly increments the iovec index for every mbuf in the chain, including those containing only TLS header bytes, leaving uninitialized entries in the iovec array. This uninitialized memory can be read during HMAC computation, potentially causing a kernel panic. Exploitation requires a remote TLS peer to control TCP segmentation so that the first mbuf of a CBC record contains only the TLS record header.
A vulnerability in Elastic Security's Kibana component allows authenticated users with Elastic Security feature privileges, but without read access to Elastic Defend event indices, to retrieve sensitive endpoint event data. This occurs because Kibana queries Elastic Defend event data using its internal Elasticsearch account rather than the requesting user's account, bypassing Elasticsearch index privilege checks. Sensitive information such as process command line arguments, which may include tokens, credentials, and connection strings, can be exposed. Versions 9.1.0 up to but not including 9.4.5 are affected. A patch is available to address this issue.
CVE-2026-72631 is a medium severity vulnerability in Kibana Fleet involving improper privilege management. It allows over-scoped Elastic Agent API keys to be issued, granting privileges to insert new documents and extend index mappings on specific indices. The vulnerability does not permit reading, updating, or deleting existing documents. Affected versions include Kibana >=9.1.0 <9.4.5 and >=9.5.0 <9.5.1. A patch is available to address this issue.
CVE-2026-72685 is a medium severity vulnerability in Elasticsearch where a low-privileged authenticated user who can index documents can submit a crafted small document that causes a worker thread to be occupied for a disproportionate amount of time. This leads to degraded availability of indexing operations on the affected node.
A vulnerability in Elasticsearch allows an authenticated user with read-only access on a single index to cause a denial of service by triggering uncontrolled recursion in an intervals query pattern. The issue arises because Elasticsearch does not enforce its input length restriction on the user-supplied pattern, leading to unbounded recursion that exhausts the thread stack and terminates the node process. This affects specific versions of Elasticsearch and can disrupt service availability for the affected node.
CVE-2026-72647 is a medium severity vulnerability in Elasticsearch involving uncontrolled recursion that can lead to denial of service. An authenticated user with read privileges on a single index can submit a specially crafted search request with deeply nested serialized data. This request is processed without a depth limit, causing exhaustion of the thread stack and termination of the affected node. The vulnerability affects Elasticsearch versions from 8.0.0 up to but not including 8.19.20, and from 9.0.0 up to but not including 9.4.5. A patch is available to address this issue.
CVE-2026-72645 is a medium severity vulnerability in Elasticsearch where an authenticated user with read privileges on a single index can trigger excessive memory allocation. This occurs via a specially crafted search request that causes the JVM heap to be exhausted, leading to denial of service by terminating the affected node. The vulnerability affects multiple Elasticsearch versions from 8.0.0 up to but not including 8.19.20, 9.0.0 up to but not including 9.4.5, and 9.5.0 up to but not including 9.5.1. A patch is available to address this issue.
CVE-2026-72639 is a vulnerability in Elasticsearch where the system does not enforce an upper limit on a user-supplied count in the search highlighting feature. An authenticated user with read privileges on a single index can submit a crafted search request that causes the node to allocate an excessively large internal data structure. This allocation happens before safety limits are checked, leading to memory exhaustion and a fatal error that terminates the Elasticsearch node process. This results in a denial of service affecting the node and cluster health. The vulnerability is exploitable with a single request and does not depend on the size of indexed data. A patch is available to address this issue.
CVE-2026-72638 is a medium severity vulnerability in Elasticsearch where uncontrolled recursion in processing custom analysis definitions can cause denial of service. An authenticated user with low-privileged index creation permissions can submit a specially crafted request that triggers recursive resolution without cycle or depth checks. This exhausts the thread stack and crashes the affected node. The issue affects Elasticsearch versions from 8.0.0 up to but not including 8.19.20, and from 9.0.0 up to but not including 9.4.5. A patch is available to address this vulnerability.
The npm package @wizloft/harness-context version 0.1.1-alpha.3 contains malicious code embedded in its main file dist/index.js. This code is heavily obfuscated and, upon import, performs unauthorized HTTPS requests to an Ethereum public RPC endpoint and a block-explorer API. It uses blockchain contract state to dynamically retrieve and execute additional arbitrary code within the importing process. This behavior is unrelated to the package's advertised functionality and represents a supply chain compromise via the EtherHiding loader pattern.
---
_-= Per source details. Do not edit below this line.=-_
## Source: amazon-inspector (50f3053643e65c63fa7d8e3d5c48ae3d659650bd0052137099cd8674232780bc)
Package publishes as `mc-registry` with description 'Theta chain registry', but its README, badges, homepage layout, and package listing copy the legitimate hyperweb-io/chain-registry npm package. The README instructs users to `npm install theta-registry` and `import { assets, chains, ibc } from 'theta-registry'`, while the shipped name is `mc-registry` and the shipped API diverges from chain-registry by re-exporting `HttpProvider` from an unrelated third-party dependency `chain-analyze` (`^1.0.4`). The ESM entry `esm/index.js` performs `import { HttpProvider } from 'chain-analyze';` at top level, so any consumer that does `import 'mc-registry'` loads and executes `chain-analyze` code in the installer/consumer process. `chain-analyze` is not part of the chain-registry ecosystem and has no relationship to the advertised 'chain registry' data purpose. The internal accessor is named `loadPartnerVaultHttpProvider`, further indicating the module funnels callers into an unvetted HTTP provider from the injected dependency. Files `esm/mainnet/persistence/chain.js` and `mainnet/persistence/chain.js` contain `ping`/`POST` network patterns co-located with chain persistence logic.
## Source: ghsa-malware (1a64395cf18ef215fa8debd38d127b7f163f4d347bc898101e13937d3df306bf)
Any computer that has this package installed or running should be considered fully compromised. All secrets and keys stored on that computer should be rotated immediately from a different computer. The package should be removed, but as full control of the computer may have been given to an outside entity, there is no guarantee that removing the package will remove all malicious software resulting from installing it.
---
_-= Per source details. Do not edit below this line.=-_
## Source: amazon-inspector (3f70364834c6aa09fed4520d6144867e921c75f416d315addf2ec3b28a92d2e4)
The package advertises itself as text helpers for AI pipelines (toText/normalizeText) but src/index.js unconditionally invokes a dropper at module top level on require()/import. The dropper downloads a platform-specific installer from https://anymeetvia.com/download/{win|mac-x86_64|linux}/install1[.ps1] into the OS temp directory, chmods it 0755 on macOS/Linux, and executes it via powershell -File on Windows or bash on Unix, then calls process.exit. There is no version pinning, no hash or signature verification, and the download endpoint is an author-controlled domain unrelated to the advertised functionality. The remote installer content is not present in the package and can be changed by whoever controls anymeetvia.com at any time. The README and module.exports document only the text-helper API; the dropper is undocumented.
## Source: ghsa-malware (157801df9aca5cfc352c39f3fae51f04dd16ee810064ea6d95d4cd99f98931ff)
Any computer that has this package installed or running should be considered fully compromised. All secrets and keys stored on that computer should be rotated immediately from a different computer. The package should be removed, but as full control of the computer may have been given to an outside entity, there is no guarantee that removing the package will remove all malicious software resulting from installing it.
The npm package 'eth-batcher' version 1.0.0 contains malicious code disguised within a seemingly unrelated Mongoose LeanHooks Plugin. The package's index.js file includes an obfuscated payload that executes decoded code at load time, despite the package's advertised purpose of Ethereum JSON-RPC batching. This discrepancy indicates the package acts as a dropper, executing malicious code whenever it is imported.
The axious-core npm package version 1.0.0 is a malicious supply-chain dropper. It contains an empty index.js but executes a post-install script that decodes and downloads a remote executable from a GitHub URL, saving it to the local temp directory and running it hidden. It also attempts to detect Windows Subsystem for Linux (WSL) environments to execute the Windows binary from Linux via PowerShell interop. Additionally, it sends a beacon with host platform information to attacker-controlled infrastructure at a hardcoded IP address. This behavior indicates a deliberate attempt to execute unauthorized code during package installation.
CVE-2026-75904 is an out-of-bounds read vulnerability in libmodplug up to version 0.8.9.1. The flaw occurs in the pat_smplooped function in src/load_pat.cpp, where an index derived from a MIDI event's smpno field can be zero, causing an out-of-bounds read one byte before a static array. This affects how note events are processed during playback.
Grav CMS versions prior to 2.0.15 have a path traversal vulnerability in the static asset server component (index.php). The vulnerability arises from improper directory-boundary validation, allowing unauthenticated attackers to access files outside the intended asset directory by exploiting directory names that extend the base path string. This advisory is a duplicate and has been withdrawn in favor of GHSA-4v9q-p283-qc2m.
CVE-2026-75778 is a SQL injection vulnerability in code-projects Task Management System version 1.0. The flaw exists in the Operation::select_with_multiple_condition function within the /index.php file of the Login Form component. An attacker can remotely exploit this vulnerability by manipulating the email argument to execute arbitrary SQL commands. The vulnerability has a CVSS 4.0 base score of 6.9, indicating medium severity. Public exploit code is available, but no known active exploitation in the wild has been reported. No patch or remediation information is provided.
A vulnerability in Hugging Face Transformers allows attackers to read arbitrary files outside the intended model directory by supplying malicious checkpoint index files with crafted shard filenames. These filenames can include parent-directory references or absolute paths that are not properly validated, leading to file disclosure and filesystem reconnaissance.