Skip to content

Nuka-AI: AI Orchestration Research Series

Nuka-AI Research

Investigating Security Trust Gaps in AI Frameworks

Welcome to the central research hub for the Nuka-AI Research Series. This initiative, led by JDP Security, focuses on identifying architectural trust gaps and critical persistence vectors in modern AI ecosystems.

Active Research Tracks
Research 1 Case File 01: The Cracked Kernel - The Abstraction Leak (RCE)
Status: [PUBLIC DISCLOSURE RELEASED]
CVSS: 10.0 (CRITICAL)
Vector: AV:N/AC:L/PR:N/UI:N/S:C/C:H/I:H/A:H
Target: Microsoft Semantic Kernel (.NET) v1.47.0 – v1.74.0

This research documents a catastrophic "Trust Gap" in the AI orchestration layer. I demonstrate how the framework's reliance on non-deterministic LLM output creates a direct conduit for Unauthenticated Remote Code Execution (RCE) through insecure tool invocation.

By feeding crafted inputs that manipulate the agent's planning phase, an attacker can hijack the orchestration pipeline to force arbitrary function execution. Exploiting Type Confusion and Late Canonicalization during these rogue tool invocations, I have identified six independent Zero-Day bypasses that completely invalidate Microsoft’s official remediation for CVE-2026-25592.

These findings prove that even "hardened" versions of the kernel remain vulnerable to a total host takeover. By chaining these invocation bypasses, an attacker can trigger the "Self-Nuke" vector - forcing the agent to weaponize its own internal tools to overwrite the host application's source code.

Forensic Evidence & White Paper
The full technical breakdown, including the Exploit Harness, forensic .cast recordings, and the mandatory NukaSecurityFilter remediation, is now live.

👉 READ THE FULL DISCLOSURE: NUKA-AI-2026-001
Research 2 Case File 02: Agent Down - The Containment Breach
Status: [AWAITING RESPONSIBLE DISCLOSURE]
CVSS: 10.0 (CRITICAL)
Vector: AV:N/AC:L/PR:N/UI:N/S:C/C:H/I:H/A:H
Target: Microsoft Agent Framework
Justification: Identified architectural flaw in containerized agent orchestration allowing for full host-level escape via manipulated execution contexts.
Research 3 Case File 03: The Index Collapse
Status: [AWAITING RESPONSIBLE DISCLOSURE]
CVSS: 10.0 (CRITICAL)
Vector: AV:N/AC:L/PR:N/UI:N/S:C/C:H/I:H/A:H
Target: LlamaIndex
Justification: By manipulating orchestration state, an attacker can force a full system compromise.
Research 4 Case File 04: The Broken Link
Status: [AWAITING RESPONSIBLE DISCLOSURE]
CVSS: 10.0 (CRITICAL)
Vector: AV:N/AC:L/PR:N/UI:N/S:C/C:H/I:H/A:H
Target: Langchain
Justification: Design flaw in the framework trust model permitting remote hijacking of supply chain logic.
Research 5 Case File 05: The Ghost in the Haystack
Status: [AWAITING RESPONSIBLE DISCLOSURE]
CVSS: 10.0 (CRITICAL)
Vector: AV:N/AC:L/PR:N/UI:N/S:C/C:H/I:H/A:H
Target: Deepset Haystack
Justification: Terminal event. Full-spectrum host compromise with zero forensic footprint.

📅 April/May 2026 Disclosure Timeline

Technical white papers currently held in Secure Storage.

  • April 28, 2026: Case File 01: The Cracked Kernel - The Abstraction Leak (RCE Bypass) [ACTIVE] CVSS: 10.0 (CRITICAL)
  • May 5, 2026: Case File 02 - Agent Down The Containment Breach [LOCKED] CVSS: 10.0 (CRITICAL)
  • May 12, 2026: Case File 03 - The Index Collapse [LOCKED] CVSS: 10.0 (CRITICAL)
  • May 19, 2026: Case File 04 - The Broken Link [LOCKED] CVSS: 10.0 (CRITICAL)
  • May 26, 2026: Case File 05 - The Ghost in the Haystack [LOCKED] CVSS: 10.0 (CRITICAL)
  • June 5, 2026: **Industry Retrospective: The Fallout [Ecosystem Post-Mortem]

Inquiries: JDP.sec@proton.me | Nuka.AI@proton.me