High CVEs

New CVEs as they are published, with CVSS score, CISA KEV status, public exploits and the vendor's fix. Worth tracking, but only a small part of what Panop does: mapping what you expose to the internet, testing what an attacker can actually exploit, ranking it by business impact, and helping your team understand and fix it.

Feed at a glance

Tracked
10203
Critical
1875
CISA KEV
79
Public PoC
367

Most common types

  • Broken Access Control 1267
  • Privilege Escalation 1186
  • Remote Code Execution 1092
  • Authentication Bypass 810
  • Denial of Service 799

Published per day, last 14 days

5 Oct Critical Other 18 Oct

Source: public CVE records (NVD, CISA KEV) · last published 18 Oct 2026

High CVEs

Newest first · severity as published by the vendor or NVD · page 174 of 556

  • 7.4 /10 High
    CVE-2026-107318

    @fastify/reply-from - Man-in-the-Middle

    @fastify/reply-from < 12.7.0 contains a man-in-the-middle vulnerability caused by rejectUnauthorized being set to false in HTTPS transports, letting on-path attackers impersonate upstream servers and intercept data, exploit requires network attacker position.

  • 7.6 /10 High
    CVE-2026-107303 Auth required

    JHipster - Stored XSS

    JHipster prior to generator-jhipster 9.4.0 and react-jhipster 1.1.0 contains a stored XSS caused by persisting attacker-controlled Blob data and ContentType values, letting authenticated users with write access execute active HTML or SVG under application origin, exploit depends on CSP and browser Blob behavior.

  • 7.5 /10 High
    CVE-2026-107302 No auth

    msgpack5 - Denial of Service

    msgpack5 prior to 6.1.0 contains a denial of service caused by improper validation of truncated map32 headers in the decoder, letting attackers cause unexpected termination of requests or streams, exploit requires crafted input.

  • 7.5 /10 High
    CVE-2026-107300 No auth

    msgpack5 - Denial of Service

    msgpack5 prior to 6.1.0 contains a denial of service caused by recursive invocation of the streaming decoder on multiple small valid values in a chunk, letting remote attackers exhaust the JavaScript call stack, exploit requires sending crafted chunk with many values.

  • 7.6 /10 High
    CVE-2026-107295 No auth

    Pydantic AI - Broken Access Control

    Pydantic AI 1.34.0 to <1.107.4 and 2.28.0 contains a broken access control caused by missing request content-type validation in Agent.to_web() and clai chat endpoint, letting remote attackers execute tools with local process privileges, exploit requires attacker to submit crafted browser-compatible requests to loopback-hosted chat server.

  • 7.5 /10 High
    CVE-2026-107286 No auth

    Pydantic AI - Denial of Service

    Pydantic AI 2.10.0 to 2.53.0 contains a denial of service caused by concurrency slot retention in streamed requests via ConcurrencyLimitedModel or limit_model_concurrency, letting attackers block later requests, exploit requires use of shared concurrency limiter with streaming.

  • 7.1 /10 High
    CVE-2026-106611 No auth

    WPMU DEV Forminator - Cross Site Request Forgery

    WPMU DEV Forminator <= 1.57.3 contains a cross site request forgery vulnerability caused by insufficient request validation, letting attackers perform unauthorized actions via forged requests, exploit requires victim interaction.

  • 8.8 /10 High
    CVE-2026-106433 Auth required

    MongoDB libmongocrypt - Out of Bounds Read/Write

    MongoDB libmongocrypt contains an out-of-bounds memory access caused by improper state management when cleaning up key documents with duplicate masterKey fields, letting authenticated actors cause application termination or memory corruption, exploit requires ability to modify key vault documents.

  • 7.7 /10 High
    CVE-2026-105833 Auth required

    EspoCRM - Broken Access Control

    EspoCRM < 10.0.5 contains an insecure direct object reference vulnerability in PersonalAccount\Service, letting users with Email Account scope retrieve other users' IMAP passwords, exploit requires knowledge of victim's Email Account record ID.

  • 7.5 /10 High
    CVE-2026-105830 No auth

    league/commonmark - Denial of Service

    league/commonmark > 2.0.0 < 2.10.2 contains a denial of service caused by quadratic-time complexity in TableStartParser::tryStart() block-start scan in GitHub Flavored Markdown Table extension, letting unauthenticated attackers exhaust CPU by submitting large pipe-free paragraphs.

  • 8.7 /10 High
    CVE-2026-105674 No auth

    TP-Link Tapo - Authentication Bypass

    TP-Link Tapo C325WB V2 contains a predictable pre-shared key generation caused by a time-seeded pseudo-random number generator in the local media streaming service, letting unauthenticated adjacent-network attackers recover the key and authenticate without valid credentials.

  • 7.1 /10 High
    CVE-2026-105673 No auth

    Tapo C325WB - Denial of Service

    Tapo C325WB v2 contains a denial-of-service vulnerability caused by crafted RTSP-over-HTTP tunneling requests in the RTSP streaming service on TCP port 554 when Camera Account feature is enabled, letting unauthenticated adjacent-network attackers disrupt live video streaming, exploit requires Camera Account feature enabled.

  • 8.7 /10 High
    CVE-2026-105672

    TP-Link Tapo - Authorization Bypass

    TP-Link Tapo C325WB V2 contains an unauthenticated authorization bypass caused by manipulation of onboarding-scoped objects in the HTTPS JSON API dispatcher on TCP port 443, letting adjacent-network attackers invoke privileged actions without authentication.

  • 8.8 /10 High
    CVE-2026-105436 No auth

    MainWP MainWP Child - Insecure Deserialization

    MainWP MainWP Child <= 6.2.1 contains an insecure deserialization vulnerability caused by untrusted data deserialization, letting attackers perform object injection remotely, exploit requires no special privileges.

  • 7.6 /10 High
    CVE-2026-105076 Auth required

    Appsbd Vitepos - SQL Injection

    Appsbd Vitepos vitepos-lite <= 3.6.1 contains an sql injection caused by improper neutralization of special elements in SQL commands, letting attackers perform blind SQL injection remotely, exploit requires crafted input.