Headline 1 of 5
Latest news
Securityobsidian-sync-mcp warns of vault access risk in unauthenticated deploymentsThe maintainers of obsidian-sync-mcp warned that a website visited while its server is running without authentication may access the Obsidian vault in versions <= 0.5.6. The advisory says the issue affects unauthenticated deployments, including Filesystem mode and CouchDB/Self-hosted LiveSync, because they lack checks that distinguish trusted local requests from browser-originated ones. The maintainers recommend upgrading to 0.5.7, which adds Host and Origin checks, or setting MCP_AUTH_TOKEN; they warn that direct non-browser clients on untrusted networks may still pose a risk.
SecuritySharpIppNext advisory flags server-side flaw in versions through 4.0.2A security advisory for SharpIppNext identifies a server-side denial-of-service flaw in versions through 4.0.2. The library buffers incoming Internet Printing Protocol document data in memory without a size limit, which can exhaust server memory. The unauthenticated, network-reachable issue affects SharpIppServer deployments; administrators should upgrade to 4.1.0 or later and apply additional safeguards.
SecurityXWiki disables REST database queries by default after data-exposure flawXWiki Platform maintainers disclosed and fixed an unauthenticated vulnerability in the XWiki REST query API that could expose database data. The flaw affected org.xwiki.platform:xwiki-platform-rest-server across three version ranges and stemmed from missing safeguards in how the API handled queries. The fixed versions are 16.10.18, 17.10.5 and 18.2.0; administrators should upgrade or temporarily block REST query functionality at the web server or reverse proxy.
SecurityServo advisory says sandboxed frames can navigate themselves automaticallyServo’s security advisory reports that attacker-controlled content in a sandboxed iframe can navigate itself automatically through declarative refresh, despite settings meant to block automatic features. The flaw affects cargoservo versions earlier than 0.3.0, which the advisory identifies as patched. The confirmed scope includes framed phishing, content changes after display and redirect loops; the advisory does not establish script execution or access to the embedding page’s DOM.
SecurityNATS Server advisory urges upgrades for MQTT-related flawA security advisory identifies a protocol-injection flaw in NATS Server versions earlier than 2.12.9 and certain 2.14 releases. The advisory says the flaw can allow arbitrary NATS commands to be injected through MQTT subscription handling and forwarding. The advisory recommends upgrading to 2.12.9 or 2.14.1; disabling anonymous MQTT access may help where MQTT is not needed.
SecurityGitHub flags code-injection flaw in Jane’s OpenAPI packageGitHub published an advisory on GHSA-xwcw-j52r-34x4 for Jane’s `jane-php/open-api-3` package. The advisory says versions earlier than 7.12.0 are affected when a generated client is used with an attacker-influenced OpenAPI document. GitHub recommends upgrading to 7.12.0 or later to address the code-injection vulnerability.
SecurityGitHub advisory flags code-execution risk in jane-php/open-api-3A GitHub security advisory warns of a code-execution risk in the Composer package jane-php/open-api-3. The advisory lists versions earlier than 7.12.0 as affected when using a client generated from an attacker-influenced OpenAPI document. Users should update to 7.12.0 or later.
SecurityJane PHP package patched for code-injection flawGitHub published advisory GHSA-q8v2-3j4h-xg2p for Jane’s `jane-php/open-api-3` Composer package. The advisory says versions earlier than 7.12.0 are affected by a code-injection vulnerability that can allow arbitrary PHP code execution when generated client code is used. Users should upgrade to 7.12.0, the version listed as patched.
SecurityNATS Server warns of MQTT flaw exposing internal QoS 2 metadataNATS Server published an advisory on Nov. 19, 2026, about an MQTT subscribe-permission bypass that can expose internal QoS 2 metadata. The advisory’s version passages conflict on the 2.12 series, while both identify v2.14.3 as fixed. The issue concerns MQTT users whose ordinary subscribe permissions would not allow access to the internal subjects; the advisory recommends upgrading to a fixed release and says deployments not using MQTT are unaffected.
SecurityMareforma advisory says invalidated claims can still reach REPLICATED statusMareforma’s advisory says invalidated claims can still be promoted to REPLICATED. It identifies version 0.3.0 and versions earlier than 0.3.8 as affected, with invalidated claims still eligible for promotion. The advisory lists 0.3.8 as patched; users can check their installed version against the affected range.
SecurityMareforma advisory says unsigned Rekor blocks can be trustedMareforma’s security advisory says its transparency_logged status trusts an unsigned Rekor block, without checking that the block is covered by a signature. The flaw affects version 0.3.0 and releases below 0.3.8, according to the advisory. Mareforma lists 0.3.8 as patched; administrators should check their installed version and update.
SecurityMareforma flaw lets unsigned claims gain REPLICATED statusMareforma disclosed an integrity flaw that can give unsigned claims REPLICATED status, weakening the status’s meaning as evidence of validation by multiple parties. The advisory says the convergence process does not enforce cryptographic signature and validator-enrollment checks, and that affected versions are >= 0.3.0 and < 0.3.7. Systems relying on REPLICATED results for trust decisions may be affected; Mareforma lists 0.3.7 as patched and recommends upgrading.
SecurityAdvisory discloses comment access flaw in djangocms-moderationA GitHub advisory warns that staff outside a moderation review could read its comments on sites using djangocms-moderation. The notice covers versions up to and including 2.3.0. It lists a patched release and recommends that site administrators install 2.4.1.
SecurityDOMShell advisory flags unreliable read-only classificationGitHub published an advisory about a permission-classification flaw in DOMShell, affecting @apireno/domshell versions before 2.0.8. The advisory says an evaluation tool was labeled read-only despite being able to perform side-effecting actions, making its permission classification and approval signals unreliable. The published scope covers versions before 2.0.8; GitHub recommends updating to 2.0.8.
SecurityElection website traffic reached about 602 million requestsCloudflare reported that traffic to state and local election websites reached approximately 602 million requests on November 3, 2026. The company said it recorded no significant distributed denial-of-service attacks across the 1,400 election-related websites it protected that day. Cloudflare also reported mitigating 392 million website-vulnerability attacks against government election websites in the 12 months before the election.
SecurityUnrealIRCd flaw could let users run commands with server authorityUnrealIRCd contains a delimiter-neutralization vulnerability that can let any user on a network with two or more linked servers inject commands executed with server authority. The flaw affects linked-server networks because injected server-to-server commands run with the authority of the server. Administrators are advised to follow vendor mitigation instructions and applicable CISA BOD 26-04 patching guidance, or discontinue use if mitigations are unavailable.
SecurityCheck Point products affected by Gaia OS privilege flawCheck Point products are affected by a vulnerability that could let an authenticated local user gain root-level command execution. The flaw affects Firewall, Multi-Domain Security Management Server, Scalable Platforms, Security Gateway and Security Management Server because Gaia OS DynamicCLI lacks adequate access controls and authentication for a local resource. CISA advises administrators to follow vendor mitigations and applicable BOD 26-04 patching guidance, assess each asset’s internet exposure and discontinue use if mitigations are unavailable.
SecurityCloudflare reports surge in AI security-tool scansCloudflare reported a global rise in scans by AI penetration-testing and orchestration frameworks, with more than 50 families observed in three days. Cloudflare said requests involving AI testing and exploitation tools had increased 30-fold since October 1. The report highlights exposed, broadly privileged credentials as a recurring concern and recommends layered defenses and tighter access controls.





