CVE Vulnerability Catalog
Translated CVE descriptions from NVD NIST - in English
Browse vulnerabilities by packageCISA KEV catalog updated: (v2026.10.14)
Weekly CVE digest
One email a week with newly published vulnerabilities worth knowing about. No account needed.
This digest covers new vulnerabilities in general, not your servers. If you want to know which of them actually run in your infrastructure, that is what Secvalis does: it scans your machines and reports only what concerns them.
A vulnerability in Dasel versions 3.0.0 through 3.10.1 causes an infinite loop when tokenizing an unterminated regex literal (e.g., r/). An attacker can exploit a crafted selector string to exhaust CPU resources.
Dasel versions 3.0.0 through 3.10.0 have a vulnerability in the parseCurRune function of the tokenizer that fails to perform bounds checking when processing escape sequences in quoted strings. An attacker can supply a crafted selector string, causing an index-out-of-range panic.
A vulnerability in PyMdown Extensions (versions 10.0.1 through 10.21.3) allows bypassing the base path restriction in the SnippetPreprocessor.get_snippet_path() function. The string-prefix containment check enables reading files from sibling directories sharing the same base_path prefix, representing a regression of the fix for CVE-2023-32309.
The Apify MCP server before version 0.9.21 has a vulnerability in the fetch-apify-docs tool where String.startsWith() is used instead of URL hostname comparison for domain validation. This allows attacker-controlled URLs to bypass the allowed domain list and return arbitrary content to the LLM.
dbt-mcp before version 1.17.1 does not sanitize node_selection and resource_type values before appending them to the dbt subprocess argument list. An MCP client can inject global dbt flags (e.g., --profiles-dir) despite shell=False.
A flaw was found in Keycloak's group search functionality in the administrative API. When Fine-Grained Admin Permissions (FGAP) v2 is enabled, a delegated administrator can bypass access restrictions to view parent groups they are not authorized to see. Searching for a child group incorrectly returns full details of the parent group, disclosing sensitive group attributes and configuration.
A vulnerability in AWS Bedrock AgentCore Python SDK versions 1.4.8 and 1.5.0 causes unintended logging of sensitive user content to CloudWatch Logs. The SDK writes raw user prompts and agent responses as OpenTelemetry span attributes without filtering or masking.
An improper access control vulnerability was discovered in Lenovo Smart Connect for Windows. A local authenticated user could access files owned by another user on the same system.
In Yamcs prior to versions 5.12.8 and 5.13.2, the PacketsApi.exportPackets endpoint failed to enforce object-level ReadPacket privileges when a request omitted specific packet names. An empty name list bypassed the WHERE pname IN filter, allowing a low-privileged or zero-privilege authenticated user to dump the entire raw telemetry packet archive.
A vulnerability in the Buffa library (Rust Protocol Buffers implementation) before version 0.8.0 causes excessive heap memory allocation when decoding unknown fields from untrusted input. An attacker can send a small, crafted protobuf payload triggering roughly 22x memory amplification (e.g., 64 MB input forcing ~1.4 GB allocation), leading to memory exhaustion and process crash.
A soundness bug in the OwnedView<V> type of the Buffa library (a pure-Rust Protocol Buffers implementation) allowed safe Rust code to trigger a use-after-free. The OwnedView::decode constructor incorrectly transmuted a borrowed slice to &'static [u8], and the Deref implementation exposed the promoted 'static lifetime on view fields, enabling access to freed memory.
Squid versions prior to 7.6 contain a heap-based buffer overflow vulnerability in cache digest reply handling. The flaw is due to improper input validation of the on-the-wire size of a cache digest, which may exceed the declared mask_size.
A vulnerability in the Claude Code Action GitHub action before version 1.0.74 allows an attacker to execute arbitrary code on the GitHub Actions runner by submitting a pull request with a malicious .mcp.json file. The action read this file from the PR branch by default and enabled all MCP servers, enabling theft of secrets (e.g., API keys and tokens) available in the workflow.
Squid versions prior to 7.6 have an out-of-bounds read vulnerability in the FTP gateway. Improper validation of directory listing date syntax may allow a trusted client to read memory from unrelated transactions.
A vulnerability in Janssen Project prior to version 2.0.0 allows acceptance of unsigned JWE request objects. This occurs because JwtAuthorizationRequest skips inner signature validation and AuthzRequestService does not reject the unrecognized RSA-OAEP algorithm.
The rz-libdemangle library (part of Rizin) contains an out-of-bounds read vulnerability in the Rust demangler (src/rust/rust_v0.c) when the demangler structure is not yet initialized. The issue is fixed in commit 6bf56d3.
In Yamcs before version 5.12.7, the authentication endpoint POST /auth/token lacks rate limiting, account lockout, or failed-attempt throttling. An unauthenticated remote attacker can perform unlimited password-guessing attempts against any user account.
In Yamcs prior to version 5.12.7, the IAM API endpoints listUsers, getUser, listGroups, and getGroup did not enforce the required SystemPrivilege.ControlAccess check. This allowed any authenticated user, even with low privileges, to enumerate all user accounts, including usernames, superuser status, and group memberships.
A missing authentication vulnerability in Windows allows a local privileged attacker to use WMI commands to arbitrarily trigger a System Management Interrupt handler. This could lead to privilege escalation or system instability.
A potential out of bounds write vulnerability could allow a local privileged attacker to execute code in System Management Mode.
Vulnerability data from NVD (NIST) · CISA KEV · EPSS

