$ ls /blog

Blog

Technical deep dives, practical lessons, personal experiences, and the occasional random thing worth sharing, all shaped by real-world experience and curiosity.

/ posts

7 posts
Abstract visualization of AI requests reusing cached context, application data, and shared knowledge to reduce repeated token processing.

Reducing AI Token Usage with Caching

Explore how prompt caching, application caching, retrieval, context management, and better prompt design can reduce token usage when building applications with OpenAI, Claude, Gemini, and Grok.

The reconnaissance agent mapping the Helixora portal, staging application, APIs, directory, and Northstar File Relay through a constrained discovery workflow.

The Reconnaissance Agent [Lab Part 4]

Follow the Helixora reconnaissance agent as it discovers staging, exposed configuration, employee information, API documentation, anonymous shares, credentials, and recovery metadata.

A security agent sending a proposed action through a restricted tool interface, scope validation, a hostname allowlist, and a local lab service.

Designing Safe Tools for Security Agents [Lab Part 3]

How the Helixora learner lab constrains AI security agents with an explicit hostname allowlist, bounded enumeration, role-specific tools, local services, human approval, and auditable evidence.

Abstract AI security lab showing a central agent connected to applications, identities, APIs, evidence, databases, and defensive monitoring systems.

Inside the Lab [Lab Part 2]

A guided tour of the fictional applications, identities, APIs, evidence, and defensive telemetry inside the self-contained AI agent reconnaissance lab.

Exploring how Microsoft Entra Conditional Access can govern access for AI agents, autonomous identities, and agent user accounts.

Conditional Access for AI Agents

Exploring how Microsoft Entra Conditional Access extends Zero Trust controls to AI agents, agent identities, autonomous access, and agent user accounts.

Exploring how AI agents can transform the reconnaissance phase of security assessments by orchestrating discovery and reasoning about the environment.

AI Agents for Security Reconnaissance [Lab Part 1]

Exploring how AI agents can transform the reconnaissance phase of security assessments by orchestrating discovery and reasoning about the environment.

Metasploit Framework 6.5 brings a collection of improvements across the framework, including authentication, Active Directory, credentials, payloads, modules, and post-exploitation capabilities.

Metasploit Framework 6.5: What's New and Why It Matters

Metasploit Framework 6.5 brings a collection of improvements across the framework, including authentication, Active Directory, credentials, payloads, modules, and post-exploitation capabilities.