#git

### Summary When vLLM is configured to use Mooncake, unsafe deserialization exposed directly over ZMQ/TCP on all network interfaces will allow attackers to execute remote code on distributed hosts. ### Details 1. Pickle deserialization vulnerabilities are [well documented](https://docs.python.org/3/library/pickle.html). 2. The [mooncake pipe](https://github.com/vllm-project/vllm/blob/9bebc9512f9340e94579b9bd69cfdc452c4d5bb0/vllm/distributed/kv_transfer/kv_pipe/mooncake_pipe.py#L206) is exposed over the network (by design to enable disaggregated prefilling across distributed environments) using ZMQ over TCP, greatly increasing exploitability. Further, the mooncake integration opens these sockets listening on all interfaces on the host, meaning it can not be configured to only use a private, trusted network. 4. The root problem is [`recv_tensor()`](https://github.com/vllm-project/vllm/blob/9bebc9512f9340e94579b9bd69cfdc452c4d5bb0/vllm/distributed/kv_transfer/kv_pipe/mooncake_pipe.py#L25...

### Impact The [outlines](https://dottxt-ai.github.io/outlines/latest/) library is one of the backends used by vLLM to support structured output (a.k.a. guided decoding). Outlines provides an optional cache for its compiled grammars on the local filesystem. This cache has been on by default in vLLM. Outlines is also available by default through the OpenAI compatible API server. The affected code in vLLM is [vllm/model_executor/guided_decoding/outlines_logits_processors.py](https://github.com/vllm-project/vllm/blob/53be4a863486d02bd96a59c674bbec23eec508f6/vllm/model_executor/guided_decoding/outlines_logits_processors.py), which unconditionally uses the cache from outlines. vLLM should have this off by default and allow administrators to opt-in due to the potential for abuse. A malicious user can send a stream of very short decoding requests with unique schemas, resulting in an addition to the cache for each request. This can result in a Denial of Service if the filesystem runs out of ...

### Summary The `fast-jwt` library does not properly validate the `iss` claim based on the RFC https://datatracker.ietf.org/doc/html/rfc7519#page-9. #### Details The `iss` (issuer) claim validation within the fast-jwt library permits an array of strings as a valid `iss` value. This design flaw enables a potential attack where a malicious actor crafts a JWT with an `iss` claim structured as `['https://attacker-domain/', 'https://valid-iss']`. Due to the permissive validation, the JWT will be deemed valid. Furthermore, if the application relies on external libraries like `get-jwks` that do not independently validate the `iss` claim, the attacker can leverage this vulnerability to forge a JWT that will be accepted by the victim application. Essentially, the attacker can insert their own domain into the `iss` array, alongside the legitimate issuer, and bypass the intended security checks. #### PoC Take a server running the following code: ```js const express = require('express') const ...

### Summary `reviewdog/action-setup@v1` was compromised March 11, 2025, between 18:42 and 20:31 UTC, with malicious code added that dumps exposed secrets to Github Actions Workflow Logs. Other reviewdog actions that use `reviewdog/action-setup@v1` would also be compromised, regardless of version or pinning method: - reviewdog/action-shellcheck - reviewdog/action-composite-template - reviewdog/action-staticcheck - reviewdog/action-ast-grep - reviewdog/action-typos ### Details Malicious commit: https://github.com/reviewdog/action-setup/commit/f0d342d fix/retag via version upgrade: https://github.com/reviewdog/action-setup/commit/3f401fe See the detailed report from Wiz Research: [Wiz Blog Post](https://www.wiz.io/blog/new-github-action-supply-chain-attack-reviewdog-action-setup) and reviewdog maintainer annoucement: [reviewdog #2079](https://github.com/reviewdog/reviewdog/issues/2079)

With financial stress at an all-time high, people are desperately seeking relief. Sadly, scammers know this all too well.

Hackers are using .VHD files to spread VenomRAT malware, bypassing security software, reveals Forcepoint X-Labs. Learn how this stealthy attack works and how to protect yourself.

Austin, TX, United States, 19th March 2025, CyberNewsWire

Sperm donor giant California Cryobank has announced it has suffered a data breach that exposed customers' personal information.

In today’s digital world, security breaches are all too common. Despite the many security tools and training programs available, identity-based attacks—like phishing, adversary-in-the-middle, and MFA bypass—remain a major challenge. Instead of accepting these risks and pouring resources into fixing problems after they occur, why not prevent attacks from happening in the first place? Our upcoming

The U.S. Cybersecurity and Infrastructure Security Agency (CISA) on Tuesday added a vulnerability linked to the supply chain compromise of the GitHub Action, tj-actions/changed-files, to its Known Exploited Vulnerabilities (KEV) catalog. The high-severity flaw, tracked as CVE-2025-30066 (CVSS score: 8.6), involves the breach of the GitHub Action to inject malicious code that enables a remote