Research

Papers, deep-dive analyses, security write-ups, and engineering guides.

Entries

2026-05-08

Do Verified Trajectories Improve Software-Agent Reliability?

Research on conditioning LLM agents with verified experience trajectories: 4-condition pilot (80+240 runs) showing modest reliability gains over instruction-only baselines

2026-05-08

Interface Reliability in Verified Software-Agent Evaluation

Paper examining whether verified trajectories improve interface reliability in software-agent evaluation tasks

2026-05-21

Prompt Optimizer: Fine-Tuning LLMs to Rewrite Vague Prompts

Qwen2.5-3B-Instruct + QLoRA pipeline that transforms vague prompts into structured, effective ones. 1183 pairs, full training and evaluation

2025-07-22

Salt Typhoon Cyberattack Analysis

Analysis of Chinese state-sponsored telecom espionage targeting nine U.S. carriers and global infrastructure

2025-07-29

SharePoint CVE-2025-53770 Analysis

Deep dive into the critical SharePoint RCE vulnerability bypass (ToolShell variant) with active exploitation

2025-07-29

Naval Group Cyber Breach Analysis

Investigation of the alleged 1TB data theft from France's leading naval defense contractor

2025-07-24

CitrixBleed 2 Vulnerability Analysis (CVE-2025-5777)

Critical session hijacking and MFA bypass in NetScaler ADC/Gateway with 11.5M+ attack attempts

2025-08-12

Spacecraft Software Security Vulnerabilities

Black Hat USA 2025 research on critical flaws in satellite command-and-control open-source software

2025-08-18

Helsinki 2024 Data Breach Investigation

OTKES investigation analysis of the KASKO breach affecting 300K+ people in Helsinki's education sector

2025-08-14

The Red 40: China's Cyber Ecosystem

Analysis of 40 elite Chinese hackers from grassroots hacktivists to state-sponsored APT architects

2025-07-22

The Conficker Worm

Technical and historical analysis of history's most restrained superworm despite infecting 9-15M systems

2026-04-10

Prompt Engineering 07: Evaluating Prompts

Building evaluation loops that catch failures before users do -- from labeled datasets to automated scoring

2026-04-10

Prompt Engineering 06: Context Engineering and Long Context

Why long context degrades accuracy, why needle-in-a-haystack is misleading, and what actually works

2026-04-10

Prompt Engineering 05: Agent Design Patterns

ReAct, Reflection, Plan-and-Execute, Tool Use, Multi-Agent Orchestration -- five patterns for reliable agents

2026-04-10

Prompt Engineering 04: Prompting for RAG Systems

Prompt-side concerns in RAG: grounding, hallucination handling, chunk relevance, and retrieval quality

2026-04-10

Prompt Engineering 03: Structured Output and Production Prompts

From chat toy to production component: JSON mode, schema enforcement, cost control, and defense against misuse

2026-04-10

Prompt Engineering 02: Reasoning and Chain of Thought

Why "just answer" fails, how CoT gives models compute budget, and advanced reasoning techniques

2026-04-10

Prompt Engineering 01: Fundamentals

Core principles of prompt engineering: clarity, specificity, structure, and iteration as the four levers

2024-12-01

Polymorphic Malware Detection

Exploring detection techniques for polymorphic malware

2024-11-15

Honeypot Implementation

Building a meta honeypot for security research

2024-10-20

VDI Performance Analysis

Diagnosing and troubleshooting VDI environments

2024-09-10

MFA Implementation Patterns

Different approaches to multi-factor authentication

2024-08-05

Compression Algorithms in Rust

Testing and comparing compression techniques

2024-07-01

Network Security Analysis

Tools and techniques for network security