Gulf Shores, Alabama · Corporate restaurant partner → Semi-truck fleet owner → AI slop king
I build AI systems that ship. I train teams to run faster with Claude Code.
AI Agent Infrastructure · Claude Code · Self-Hosted Infrastructure · Production SaaS
Building Now · Live in Production · Open Source · Contributions · Philosophy · Connect
| Project | What it is |
|---|---|
| Claude Code Plugins | The Claude Code plugin + skill marketplace (tonsofskills.com) — 400+ plugins, 2,900+ skills, 17 packs · 2,500★ |
| Intent Eval Platform | Vendor-neutral eval lab — 6 composable Apache-2.0 repos converging on one signed Evidence Bundle (in-toto → Rekor). Measure the behavior, sign the result, gate the ship |
| Intentional Cognition OS + Intent KB | The Compile-Then-Govern brain — facts compiled into governed, audited memories, then queried with citations. The full compile → govern → search loop runs green in production |
| Braves Booth | Live broadcast ops dashboard for Atlanta Braves radio — stats-enriched AI narratives, in production since April 2026, self-hosted |
| DiagnosticPro | AI diagnostic reports for vehicles and equipment — $4.99 per report or $29/mo. Fully self-hosted, powered by OpenAI gpt-4o |
| Claude Code Slack Channel | Slack-native governance substrate for Claude Code — Ed25519-signed audit journal, 1,251 tests · 35★ |
| Guidewire MCP | Governed carrier-native MCP servers for Guidewire InsuranceSuite — PolicyCenter, ClaimCenter, BillingCenter (v0.1.1, 135 tests) |
| Agent Governance Plane | Sandbox + governance plane for AI coding agents — sandboxed execution, human-in-the-loop approvals, signed audit log |
| CAD AI Agent | Drawing intelligence for AEC — upload a DXF/PDF/DWG, describe the change in plain English, get structured edits and compliance reports without the AI touching your original file |
Also in flight: intent-outreach (Claude-Code-native SDR orchestrator) · contributing-clanker
Everything below runs on one self-hosted Linux VPS — single Caddy ingress, Docker, systemd. No cloud vendor.
- ⚾ scorecardecho.com — Braves Booth broadcast dashboard, live on air since April 2026
- 🔧 diagnosticpro.io — AI diagnostic reports, $4.99 + $29/mo
- ⚽ hustlestats.io — youth-sports analytics PWA
- 🧩 tonsofskills.com — the Claude Code plugin + skill marketplace
- 🏗️ Internal stack, same box — Plane (projects), ERPNext (ERP), Twenty (CRM), Umami (analytics)
| Project | Description | |
|---|---|---|
 |
Claude Code Plugins | 400+ plugins, 2,900+ skills, 17 packs — the largest Claude Code plugin ecosystem |
 |
Intent Eval Platform | Audit-first AI evaluation — 6 Apache-2.0 repos, one signed Evidence Bundle |
 |
Claude Code Slack Channel | Two-way Slack bridge for Claude Code — Socket Mode + MCP stdio, tamper-evident audit journal |
 |
Intentional Cognition OS | Governance kernel for compiled machine reasoning — the Compile-Then-Govern engine |
 |
Intent Knowledge Base | Governed team memory for Claude Code — wiki-link resolution, freshness-aware search, export gating |
 |
Guidewire MCP | Carrier-native MCP + governance harness for Guidewire estates |
 |
CAD AI Agent | Drawing Intelligence Platform for AEC — non-destructive LLM edits on DXF/PDF/DWG |
Past upstream contributions across the ecosystem:
- Vertex AI Samples — ADK inline source deployment tutorial for Agent Engine
- Agent Starter Pack — Production ADK reference added to Google's community showcase
- Kilo Code — Agent runtime test fix for VSCode applyEdit
- PostHog — Immutable state updates and feature flag schedule fixes (2 PRs merged)
- tldraw — Prettier workspace extension configuration
- Filament PHP — Dark mode input fix for Tailwind CSS forms
- Blur Extension — Pattern builder UI, settings export/import, Teams and Slack screen share detection (4 PRs merged)
- Beads — Natural-language skill activation system (823 lines)
Important
The bottleneck in software isn't writing code — it's knowing what to build and shipping it before the window closes. I spent 20 years running restaurants, trucking fleets, and military operations before I wrote my first line of code. That operator's bias never left: every system I build has to survive contact with real users, real money, and real deadlines. Claude Code and modern AI are force multipliers, but only if you treat them like tools instead of magic. Ship to production. Measure what matters. Delete what doesn't work.
- Marine (USMC) and Citadel alumnus — 20+ years of operations leadership
- Self-taught developer → AI systems architect
- Self-hosted stack: Linux VPS, Caddy, Docker, systemd
- Founder of Intent Solutions — AI consulting and team training
- Gulf Shores, Alabama