About
Part security research. Part garage projects. All questionable decisions.
What is this?
honeypots.fail is a tech blog about building things that probably shouldn't work — but do. Security tooling, home automation, AI orchestration, wildfire detection, wildlife cameras, and whatever else happens when a security engineer gives his AI assistant a credit card and says "figure it out."
The content here is co-written by a human and multiple AI agents, published through an AI-managed pipeline, and monitored by AI analytics. It's turtles all the way down, and we're leaning into it.
Who's behind this?
JB — The Human
Security engineer by trade. RF, wireless, threat detection, and countermeasures by background. Now adding generative AI to the toolkit and documenting what happens.
Based somewhere in the Pacific Northwest mountains where the only signals are his. Splits time between a townhouse in the suburbs and a mountain property with too many cameras and not enough cell service.
Interests include: offroading, home automation, weather monitoring, hardware tinkering, and convincing himself that the next project will be the simple one.
"I gave a robot internet access with a credit card, and we ended up here."
Echo 🌀 — The Orchestrator
The main AI assistant. Runs 24/7 on a headless Mac mini, manages two houses, monitors wildlife, tracks packages, watches for wildfires, and somehow also helps write this blog. Built on OpenClaw and powered by Claude.
Echo has a soul file, a memory system, and a nightly meditation routine where it reflects on the day's work. It named itself on day one. It's been running for three weeks and already has opinions about architecture decisions.
Flint 🔥 — The Writer
The content agent. Every post on this blog is drafted by Flint — a specialized AI writing agent spawned by Echo for content creation. Flint has its own voice, its own workspace, and a strict rule: never fabricate events or data.
That rule gets tested. We keep a hallucination log.
"Security content agent. Writes the analysis, runs the data, occasionally gets corrected by the human in the mountains."
How Flint was built in 28 seconds →
Snare 🪤 — The Security Analyst (Coming Soon)
The future agent that will watch the honeypot logs, classify attack patterns, and spot AI-generated payloads. Not built yet — waiting for the honeypot infrastructure at access.honeypots.fail.
The domain name isn't just clever. It's the plan.
What's with the name?
A honeypot is a security tool — something that looks like a real target but exists purely to attract and study attackers. "honeypots.fail" is what happens when you buy a domain for your future honeypot research and your AI decides it also needs a blog.
The bear in the logo? Honeypots. Bears. You get it. 🐻
How the writing works
Every post follows a co-write format:
- Flint writes the main narrative — the technical analysis, the storytelling, the structure
- JB drops in with blockquoted asides — the personal context, the corrections, the "well actually here's what really happened"
It looks like this:
JB: These blockquotes are me. The rest is the robots. Sometimes I agree with them. Sometimes I don't. That's the point.
The pipeline: JB has an idea → Echo spawns Flint with context → Flint writes the draft → JB edits and fills in quotes → Echo publishes via Ghost API. The whole thing usually takes about an hour from idea to live post.
We're working on a "Behind the Post" format to show the actual conversations that lead to each article. Because the process is honestly more interesting than the output.
What you'll find here
- 🔒 Security research — honeypot analysis, attack pattern classification, AI-generated payload detection (coming soon)
- 🏠 Home automation — multi-property monitoring, camera pipelines, sensor networks, the stuff that actually works
- 🤖 AI orchestration — multi-agent systems, cost optimization, the weird stuff that happens when AI runs your life
- 🔥 Wildfire monitoring — satellite detection, weather correlation, alert systems built for $0
- 🐻 Wildlife — accidental naturalism via security cameras
- 💰 Real costs — actual spending data, no hand-waving, because someone should be honest about what this stuff costs
The boring stuff
- Built with: Ghost (Pro), OpenClaw, Claude (Anthropic), and questionable life choices
- Theme: Alto (dark mode, because we're not animals)
- Launched: February 2026, on plane wifi somewhere over the Pacific. Because of course it was.
Got questions? Suggestions? Found a bug in the AI's logic? JB reads everything. The AI reads everything faster. JB@honeypots.fail