CVE-2026-108698 is a high-severity vulnerability in hyper-mcp through version 0.8.3. It involves a time-of-check time-of-use (TOCTOU) race condition in the load_wasm function that leads to a signature verification bypass. The vulnerability allows attackers who control registry responses for a tag to serve an unsigned malicious manifest to the loader while presenting a signed manifest to the Cosign verifier. This results in execution of unsigned WebAssembly plugins with configured host capabilities.
Miniflux versions 2.3.0 through 2.3.3 have a server-side request forgery (SSRF) vulnerability that allows authenticated users to set a feed's proxy_url to internal or loopback addresses. This bypasses network restrictions intended to prevent access to private networks, enabling potential probing of internal ports and proxy-style requests to internal services. The vulnerability has a medium severity with a CVSS score of 4.3.
JupyterHub versions through 6.0.1 have an identifier collision vulnerability that allows authenticated users to overwrite another user's named-server OAuth client by registering a hyphenated username. This can disrupt OAuth login and revoke tokens for the targeted user's server by stopping the attacker's own server.
CVE-2026-108681 is a denial of service vulnerability in zhayujie CowAgent versions 2.1.0 through 2.1.7. The flaw exists in an unknown function within the file channel/web/web_channel.py of the Web Console component, where manipulation of the session_id argument can cause service disruption. Version 2.1.7 includes a partial mitigation by limiting the request body size to 512MB. The vulnerability can be exploited remotely, and public exploit code is available.
CVE-2026-108584 is a medium severity vulnerability in FunnyWolf Viper versions 3.1.0 through 3.1.11. It involves hard-coded credentials located in the file /root/viper/.git/config. This flaw allows remote attackers to exploit the system without authentication. Public exploit code is available, increasing the risk of attacks.
CVE-2026-108578 is an information disclosure vulnerability in the Neterbit NW-431F device, specifically related to an unknown function handling the /sms.json file in its Embedded Web Server component. The vulnerability can be exploited remotely without authentication. The vendor has not responded to disclosure attempts, and no patch or remediation information is currently available.
A path traversal vulnerability exists in InstantSoft icms2 up to version 2.18.2 in the Image Handler component's files_delete_file function. This vulnerability can be exploited remotely by manipulating the 'size' argument. The vulnerability has a low severity score and does not impact confidentiality or integrity but may cause availability issues. A patch has been identified and is recommended to fix the issue.
A SQL injection vulnerability exists in Furion .NET Framework up to version 4.9.9.92. The flaw is in the String.Replace function within the file framework/Furion/Templates/Extensions/StringRenderExtensions.cs. This vulnerability allows remote attackers to manipulate the argument 'Name' to perform SQL injection. Exploit code is publicly available. The vendor has not responded to disclosure attempts and no patch information is provided.
A vulnerability exists in BerriAI LiteLLM versions up to 1.94.0 in the get_secret function of secret_managers/main.py. This flaw involves improper authorization due to manipulation of the api_key argument, allowing remote attackers to exploit it. The vulnerability has a CVSS score of 6.3, indicating a moderate impact on confidentiality, integrity, and availability. The vendor was notified but has not responded, and no patch or mitigation guidance is currently available. Public exploit details have been disclosed, but no known exploitation in the wild has been reported.
A cross-site scripting (XSS) vulnerability exists in InstantSoft icms2 up to version 2.18.2 in the Private Message Handler component. The vulnerability is due to improper handling of the nickname argument in the index.tpl.php file. This flaw can be exploited remotely and has been publicly disclosed. The vulnerability has a low severity rating with a CVSS score of 3.5. A patch identified by commit 3a1ec8fcb073a46d06a2ab83bc2bf68225834281 is available to address this issue. It is recommended to apply this patch to mitigate the vulnerability.
A vulnerability in BerriAI LiteLLM up to version 1.95.0 allows remote attackers to bypass authorization by manipulating the session_id argument in the ui_view_session_spend_logs function of the Spend Tracking component. This issue is resolved by upgrading to version 1.96.0.
CVE-2026-108577 is a medium severity vulnerability in Konstanty Bialkowski's libmodplug library versions 0.8.9.0 and 0.8.9.1. It affects the ABC Music Format Parser component, specifically the abc_add_gchord function in src/load_abc.cpp. The vulnerability allows remote attackers to cause resource consumption, potentially leading to denial of service conditions. No vendor response or patch is currently available.
LlamaFarm versions up to 0.0.34 have an insecure default configuration where the unauthenticated FastAPI server binds to 0.0.0.0 on port 14345. The command-line interface (CLI) does not allow overriding the host, resulting in exposure to network-adjacent attackers. These attackers can access the project and dataset management API to read stored provider API keys, modify projects, trigger ingestion processes, and irreversibly delete projects.
mistral.rs versions 0.9.0 through 0.9.4 contain a vulnerability in the mistralrs-code-exec component where improper link resolution before file access allows sandboxed shell code to read and overwrite files outside the sandbox via symbolic links. This can be exploited by attackers or prompt-injected agents who can manipulate symlinked input or output paths to access files with the server process's permissions.
Easy!Appointments versions up to 1.6.0 have an authorization bypass vulnerability in the Booking::register() function. This flaw allows unauthenticated attackers to modify any appointment by supplying an appointment ID without the required hash. Attackers can enumerate appointment IDs sequentially, use a self-asserted manage_mode flag to alter appointment details, reassign appointments to attacker-controlled customers, and obtain management hashes for rescheduling or cancellation.
Nexting pinclaw OpenClaw channel plugin versions up to 0.3.0 contains a missing authentication vulnerability in its HTTP router. This flaw allows unauthenticated attackers to bypass the authToken check on the POST /pinclaw/send endpoint. The service listens on port 18790 bound to all interfaces by default, enabling remote attackers to inject blind prompts into the main OpenClaw agent session as user instructions.
Abilityai Trinity versions up to 0.9.5 have a missing authorization vulnerability in the Telegram router. This flaw allows agent-scoped MCP API keys to perform operations intended only for human users, such as sending messages through the owner's bot token, replacing the binding with another token, or deleting it. The vulnerability can be exploited by attackers controlling an agent, often via prompt injection.
Hatchet versions up to 0.110.5 have a vulnerability where unauthenticated attackers can exhaust server memory by sending large or concurrent requests to the SNS ingestion endpoint. This occurs because the service buffers request bodies before verifying signatures, leading to degraded availability.
CVE-2026-108754 is a vulnerability in tbphp gpt-load up to version 1.4.11 where sensitive client proxy keys are logged in cleartext. This occurs because the Gin Logger middleware logs the raw query string before the authentication key is stripped, exposing proxy keys in console logs or app.log files. Attackers with read access to these logs can recover proxy keys and misuse them against the associated group. The vulnerability has a medium severity with a CVSS score of 4.8.