Latest News

### Summary Provided grammar, would fit in a context window of most of the models, but takes minutes to process in 0.1.23. In testing with 0.1.16 the parser worked fine so this seems to be a regression caused by Earley parser. ### Details Full reproducer provider in the POC section. The resulting grammar is around 70k tokens, and the grammar parsing itself (with the models I checked) was significantly longer than LLM processing itself, meaning this can be used to DOS model providers. ### Patch This problem is caused by the grammar optimizer introduced in v0.1.23 being too slow. It only happens for very large grammars (>100k characters), like the below one. v0.1.24 solved this problem by optimizing the speed of the grammar optimizer and disable some slow optimization for large grammars. Thanks to @Seven-Streams ### PoC ``` import string import random def enum_schema(size=10000,str_len=10): enum = {"enum": ["".join(random.choices(string.ascii_uppercase, k=str_len)) for _ in...

Hello Kubernetes Community, A security issue was discovered in secrets-store-sync-controller where an actor with access to the controller logs could observe service account tokens. These tokens could then potentially be exchanged with external cloud providers to access secrets stored in cloud vault solutions. Tokens are only logged when there is a specific error marshaling the `parameters` sent to the providers. ### Am I vulnerable? To check if tokens are being logged, examine the manager container log: ```bash kubectl logs -l 'app.kubernetes.io/part-of=secrets-store-sync-controller' -c manager -f | grep --line-buffered "csi.storage.k8s.io/serviceAccount.tokens" ``` ### Affected Versions - secrets-store-sync-controller < v0.0.2 ### How do I mitigate this vulnerability? Upgrade to secrets-store-sync-controller v0.0.2+ ### Fixed Versions - secrets-store-sync-controller >= v0.0.2 ### Detection Examine cloud provider logs for unexpected token exchanges, as well as unexpected...

### Impact **What kind of vulnerability is it?** This is a **Critical** severity directory traversal (path traversal) vulnerability in the `File.download()` method of the `internetarchive` library. **Who is impacted?** All users of the `internetarchive` library versions `< 5.5.1` are impacted. The vulnerability is particularly critical for users on **Windows systems**, but all operating systems are affected. **Description of the vulnerability:** The vulnerability existed because the `file.download()` method did not properly sanitize user-supplied filenames or validate the final download path. A maliciously crafted filename could contain path traversal sequences (e.g., `../../../../windows/system32/file.txt`) or illegal characters that, when processed, would cause the file to be written outside of the intended target directory. **Potential Impact:** An attacker could potentially overwrite critical system files or application configuration files, leading to a denial of service, privil...

### Impact When establishing a TLS session using `fs2-io` on the JVM using the `fs2.io.net.tls` package, if one side of the connection shuts down write while the peer side is awaiting more data to progress the TLS handshake, the peer side will spin loop on the socket read, fully utilizing a CPU. This CPU is consumed until the overall connection is closed. This could be used as a denial of service attack on an fs2-io powered server -- for example, by opening many connections and putting them in a half-shutdown state. Note: this issue impacts ember backed http4s servers with HTTPS as a result of ember using fs2's TLS support. ### Patches Fixed in fs2 3.12.2 and 3.13.0-M7. ### Workarounds No workarounds. ### For more information If you have any questions or comments about this advisory: [Open an issue.](https://github.com/typelevel/fs2/issues/new/choose) Contact the [Typelevel Security Team](https://github.com/typelevel/.github/blob/main/SECURITY.md).

Experts see subtle improvements from new laws and best practices, but much work remains.

## Summary Insecure session handling opened room for a privilege escalation scenario in which [prebuilt workspaces](https://coder.com/docs/admin/templates/extending-templates/prebuilt-workspaces) could be compromised by abusing a shared system identity. ## Details Coder automatically generates a session token for a user when a workspace is started. It is automatically exposed via [`coder_workspace_owner.session_token`](https://registry.terraform.io/providers/coder/coder/latest/docs/data-sources/workspace_owner#session_token-1). Prebuilt workspaces are initially owned by a built-in `prebuilds` system user. When a prebuilt workspace is claimed, a new session token is generated for the user that claimed the workspace, but the previous session token for the `prebuilds` user was not expired. Any Coder workspace templates that persist this automatically generated session token are potentially impacted. For example, the [coder-login module](https://github.com/coder/registry/blob/8677e7...

Exploitation of CVE-2025-42957 requires "minimal effort" and can result in a complete compromise of the SAP system and host OS, according to researchers.

**Reporter:** Lumina Mescuwa **Product:** ImageMagick 7 (MagickCore) **Component:** `MagickCore/blob.c` (Blob I/O - BlobStream) **Tested:** 7.1.2-0 (source tag) and 7.1.2-1 (Homebrew), macOS arm64, clang-17, Q16-HDRI **Impact:** Heap out-of-bounds **WRITE** (attacker-controlled bytes at attacker-chosen offset) → memory corruption; potential code execution --- ## Executive Summary For memory-backed blobs (**BlobStream**), [`SeekBlob()`](https://github.com/ImageMagick/ImageMagick/blob/3fcd081c0278427fc0e8ac40ef75c0a1537792f7/MagickCore/blob.c#L5106-L5134) permits advancing the stream **offset** beyond the current end without increasing capacity. The subsequent [`WriteBlob()`](https://github.com/ImageMagick/ImageMagick/blob/3fcd081c0278427fc0e8ac40ef75c0a1537792f7/MagickCore/blob.c#L5915-L5938) then expands by **`quantum + length`** (amortized) instead of **`offset + length`**, and copies to `data + offset`. When `offset ≫ extent`, the copy targets memory beyond the allocatio...

The biggest vulnerabilities may lie at the boundaries of where the AI agent connects with the enterprise system.

pgAdmin <= 9.7 is affected by a Cross-Origin Opener Policy (COOP) vulnerability. This vulnerability allows an attacker to manipulate the OAuth flow, potentially leading to unauthorised account access, account takeover, data breaches, and privilege escalation.