Kali Linux (Product): news timeline & CVEs

Decepticon – Autonomous Red Team Agent that Integrates Claude and Codex

Open-source Decepticon red team agent links Claude Code and Codex to a LangGraph runtime for scoped attack-path testing.

Decepticon is an Apache 2.0 autonomous red team agent that lets Claude Code and OpenAI Codex monitor and steer engagements over the Model Context Protocol. Offensive work runs in a LangGraph server inside a Kali Linux sandbox, with 16 specialist agents covering discovery, exploitation, Active Directory, cloud, smart contracts, and reverse engineering. Developers report 102 of 104 XBOW benchmark challenges passed, though the tests included vulnerability-category hints and are not independently verified on enterprise networks. Written permission is required before testing, and a planned defensive retest loop is not yet implemented.

Cyber Security NewsOct 12, 08:05 UTC · 4d agoTools7· 3 reads

KaliBench: A Fine-Grained Benchmark for Cybersecurity Tool Use on Kali Linux with Runtime-Free Verifiable Rewards

KaliBench shows no open-weight model exceeds 42% exact-command accuracy generating Kali Linux CLI commands without tool hints.

KaliBench is a fine-grained benchmark of 8,504 natural-language-to-command pairs covering 1,642 Kali Linux tools, 23 capability dimensions, and five security phases. It uses manuscript-grounded construction, deterministic canonicalization, alias-aware scoring, and verification via LLM checks, sandboxed execution, and human review, and it supplies runtime-free verifiable rewards for training. Across 24 configurations of general-purpose and security-focused open-weight models, none exceeded 42% exact-command accuracy in the unrestricted setting. Supervised fine-tuning and reinforcement learning with KaliBench rewards substantially improved an 8B model to roughly match a 685B mixture-of-experts model.

arXiv cs.CROct 1, 17:59 UTC · 15d agoResearch2

KaliBench: A Fine-Grained Benchmark for Cybersecurity Tool Use on Kali Linux with Runtime-Free Verifiable Rewards

KaliBench measures whether LLMs can turn analyst requests into exact Kali Linux commands across 1,642 tools.

KaliBench is a benchmark of 8,504 natural-language-to-CLI pairs spanning 1,642 Kali Linux tools, 23 capability dimensions, and five security phases. Deterministic canonicalization, alias-aware scoring, LLM validation, sandboxed execution, and human review support runtime-free verifiable training rewards. Across 24 configurations, no open-weight model exceeded 42% exact-command accuracy without tool hints. Supervised fine-tuning and reinforcement learning with KaliBench rewards brought an 8B model to performance comparable to a 685B mixture-of-experts model.

Hugging Face daily papersSep 30, 20:00 UTC · 16d agoAI research

Kali Linux

Timeline3 articles

Appears with

Decepticon – Autonomous Red Team Agent that Integrates Claude and Codex

KaliBench: A Fine-Grained Benchmark for Cybersecurity Tool Use on Kali Linux with Runtime-Free Verifiable Rewards

KaliBench: A Fine-Grained Benchmark for Cybersecurity Tool Use on Kali Linux with Runtime-Free Verifiable Rewards