HarrisonSec_
HarrisonSec — videos with chapters, companion notes, and projects you can reuse.
Start with the format you prefer, then pick a track when you want to go deeper.
[ Choose_your_track ]
AI Operator
Operate AI. Ship results. Practical workflows with safety boundaries.
- Outcome-driven tutorials & walkthroughs
- Reliability + privacy/security guardrails
- Templates, checklists, failure modes
HarrisonSecurityLab
Systems internals, security research, and hard technical judgment.
- Systems internals + performance + distributed systems (latency, throughput, failure modes)
- Exploit research & reverse engineering deep dives
- Writeups + demos + tooling you can reuse
[ Latest ]
Latest Videos
Start with the most recent uploads (with chapters).
- Why Your AI Agent Keeps Failing — The 90% Problem
- The 90% AI Agent Problem (Podcast Episode)
- Why Claude Code's Agent Loop Is 1,421 Lines
Latest Notes
Text + code walkthroughs that pair with videos and projects.
- Why Hidden Data Dies in Chat Apps: LSB vs a Robust Watermark, Measured
- Pi Ships Zero Lines of MCP. Codex Ships 53,713.
- Your Agent Session Is a Tree. Your Code Thinks It Is a List.
[ Featured_Projects ]
WaxSeal: Robust Image Watermarking for Secret Delivery
Delivering a secret through chat apps that recompress every photo. Why LSB cannot survive the trip, what a robust watermark costs, and the measurements behind a two-tier design.
Secure VLAN Design: Segmentation, ACLs and Zero Trust
How an enterprise LAN gets segmented properly — the VLAN architecture, the access-control rules that enforce it, the attacks it stops, and how it is deployed and monitored.