Skip to content
KitploitKITPLOIT
ToolsExploitsBlog
Log in
Submit
ToolsExploitsBlog
Submit

Hacking, PenTest, and Cybersecurity Tools for Your Security Arsenal!

Kitploit is a directory of hacking, cybersecurity, and pentesting tools. Discover the latest project updates to find vulnerabilities, analyze systems, automate testing, and strengthen your security.

FeedsContactPrivacy© 2026 Kitploit

Tool Directory

Categories

View all categories
Loading categories
Tools/GitHubGitHub/vladimir-tokarev-cyera/llama-cpp-security-patches
Defensive ToolsVulnerability AnalysisSupply Chain SecurityAI Security
GitHubvladimir-tokarev-cyera/llama-cpp-security-patches

llama-cpp-security-patches

Security patches for unpatched llama.cpp vulnerabilities (CVE-2026-43626 through CVE-2026-43632) — Cyera Research

View Repository
Share
2663 months agoNot yet reviewed

llama.cpp Security Patches

Security patches for unpatched vulnerabilities in llama.cpp, discovered by Cyera Research.

Background

Between July 2025 and June 2026, we reported 10 vulnerabilities to the llama.cpp project through GitHub Security Advisories and MITRE. All advisories were closed by the maintainer without fixes or CVE assignment. We submitted a public PR with patches in June 2026 — it was also closed without merging.

VulnCheck allocated 10 CVE IDs after both the vendor and MITRE failed to act. Three of the original IDs (CVE-2026-43623, CVE-2026-43624, CVE-2026-43626) were subsequently replaced by CVE-2026-70638, CVE-2026-70639, and CVE-2026-70640 respectively.

These patches are provided so that users and downstream projects can protect themselves.

Unpatched Vulnerabilities

CVECVSSVulnerabilityAffected File(s)Patch
CVE-2026-706407.3Android JNI race condition (UAF)ai_chat.cppCVE-2026-70640
CVE-2026-436278.5llama_batch_init() integer overflowllama-batch.cppCVE-2026-43627
CVE-2026-436288.5DRY sampler integer underflowllama-sampler.cpp, arg.cpp, server-task.cppCVE-2026-43628
CVE-2026-436299.2KV cache state buffer overflowllama-kv-cache.cppCVE-2026-43629
CVE-2026-436319.2Server sleep-idle use-after-freeserver-context.cpp, server-queue.cpp, server-queue.hCVE-2026-43631
CVE-2026-436329.2Server unbounded task creationserver-task.cpp, server-queue.cppCVE-2026-43632

Applying Patches

Apply a single fix:

cd /path/to/llama.cpp
git apply /path/to/patches/CVE-2026-43631-server-sleep-uaf.patch

Apply all fixes at once:

cd /path/to/llama.cpp
git apply /path/to/patches/all-fixes.patch

If git apply fails due to context changes in newer llama.cpp versions, try with --3way:

git apply --3way /path/to/patches/all-fixes.patch

Disclosure Timeline

DateEvent
July 2025First vulnerability submitted via GitHub Security Advisory
October 2025Contacted MITRE for CVE assignment — 7 request IDs received, none assigned
January–May 2026Four more advisories submitted — all closed without fixes
May 2026VulnCheck allocated 10 CVEs
June 2026Public PR with patches submitted — PR closed

Researchers

  • Vladimir Tokarev (@G1ND1L4) — Vulnerability Research Tech Lead, Cyera
  • Ofek Itach (@ofekitach) — Security Research Team Lead, Cyera

Related

  • DEF CON 34 Talk: Breaking Local AI Runtimes
  • Original PR with patches (closed)
Download Tool

Most Popular

View all →

Discover the most used tools by our community.

  • 1

    OsintHunter

    letox4/osinthunter

    🎯 The ultimate OSINT toolkit: investigation graph, confidence scoring, alias families, map, timeline and reports. 49 modules, 400+ sites, no API keys required.

    938
  • 2

    ghostlock-app

    yukonga/ghostlock-app

    GhostLock One-Tap Execution App (CVE-2026-43499)

    556
  • 3

    redforge

    cybersecurity91/redforge

    Browser-side red team toolkit with 102 offline tools for recon, payload generation, credential attacks, crypto, encoding and engagement reporting.

    352
  • 4

    ARTEX

    baarai/artex

    LLM multi-agent autonomous penetration testing platform with planner/worker agents, asset and exploration graphs, MITM traffic recording, and vulnerability retesting.

    343
  • 5

    looking-glass

    anonghosty/looking-glass

    CSIRT platform that ingests Shadowserver reports into MongoDB and replays them as interactive maps, analysis, findings, advisories and executive dashboards.

    336
  • 6

    CyberStrikeAI

    ed1s0nz/cyberstrikeai

    The system of action for AI-native cybersecurity—where intent becomes governed execution, evidence becomes operational memory, and every operation improves the next.

    291

Explore all tools

Browse our collection of tools

View all tools →