How I Work

Claude Code is my office suite. It's where I write code, manage servers, deploy websites, run security audits, and publish this site. Not through chat conversations, but through a system of specialized agents, custom MCP servers, and self-hosted infrastructure that all connect through one terminal.

Everything on this page is real, running in production, and used daily. Follow the build log →

Latest from the Build Log

Three days deep on mcp-unifi. Started Wednesday with the new UCG-Fiber going live and the server flipping out of stub mode against real hardware for the first time. Shipped two release candidates, then v0.5.0, then v0.5.1. Network module split into 10 files, Protect module added (12 tools), audit lo...

Read full post →

Pinned Projects

What I've pinned on GitHub. MCP servers, tools, and apps I've built and open-sourced.

Strava MCP Vault

Strava MCP server with SQLite caching, automatic token refresh, and rate limit awareness. Built with FastMCP.

PythonFastMCPDockerSQLite

Open Model Arena

Blind AI model comparison arena with ELO leaderboard.

PythonFastAPIDocker

Claude Code Statusline

Labeled-row status bar for the Claude Code TUI. Context usage, billing tier, git status, weather, and session metrics.

ShellClaude Code

Claude Research Agent

A Claude Code skill that produces citation-grounded research reports with strict anti-hallucination rules.

Claude CodeMarkdown

MCP SearXNG

SearXNG MCP server for Claude Code. Web search, news, deep multi-page search, and people lookup via self-hosted SearXNG.

PythonFastMCPDocker

MCP Threatintel

Threat intelligence MCP server for Claude Code. IOC lookups, CVE checks, breach monitoring, dark web search, and OTX pulses.

PythonFastMCPDocker

Agents I've Designed

Specialized Claude Code agents I've built with automatic routing. Each owns a domain, has its own tools, and follows strict operational contracts.

Infrastructure

Tank Homelab operator. Containers, NAS, Docker, networking.
Sentinel Health monitor and security auditor. SSL, ports, headers.
Switch Claude Code config, LiteLLM gateway, API routing.
Forge MCP server builder. Designs, deploys, and manages MCP tools.
Keeper Production server operator. WordPress, Apache, WooCommerce.

Business

Morpheus Business ops commander. Pipeline, publishing, strategy.
Radar SEO and brand auditor. Scans sites, builds reports, deploys.
Outreach Prospect email lifecycle. Drafts, tracks, follows up.
Vet People due diligence. Public records, risk-rated dossiers.
Oracle Career strategist. Resume, positioning, org change.

Development

Link Webapps operator. Builds, deploys, and debugs applications.
Architect Workflow and system design. N8N builder.
Conduit Cornell integrations. Entra apps, Graph API, admin consent.
Excalidraw Diagrams and visual flows on self-hosted canvas.

Intelligence

Research Deep research with anti-hallucination guardrails. Citations, confidence levels.
Watch Incident monitor. Polls sources, escalates on new developments.
LinkedIn Content drafter. Builder voice, platform formatting.

Apps I've Shipped

Self-hosted applications I've built and deployed from the terminal.

The Stack

This site. Astro microblog, one-command publish from the terminal. No CMS, no database.

AstroMarkdownrsync

Phantom Paste

Zero-knowledge pastebin. Client-side encryption, auto-expiring links, no server-side plaintext.

GoSQLiteDocker

How I've Built It

Patterns and approaches I use behind the system.

5-Layer Prompt Injection Defense

Global data boundaries, per-agent hardening, two-pass analysis, canary verification, centralized logging.

Read the build log →

Anti-Hallucination Research Agent

Permission to say "I don't know." Direct quote extraction, inline citations, confidence levels, sourced reports.

Read the build log →

Agent Ecosystem Audit

20 agents audited against Anthropic guidelines. Progressive context disclosure, pipeline handoffs, error contracts.

Read the build log →

Docker-First Infrastructure

29+ containers in isolated Compose stacks. Caddy reverse proxy, Tailscale mesh, automated backups.

Push vs. Pull Prompting

Two prompting methodologies for different contexts. Explicit direction vs. expert-driven information gathering.

Read the build log →