The exact AI stack behind two 7‑figure companies — given away, free.
The exact setups I run in production — prompts, MCPs, automations — written up so you can copy them, wire them, ship them. Every guide unlocks with an email.
27 guides · all live · grouped, searchable
Claude Plugins Directory: Stop Wiring MCP By Hand
Anthropic's own curated catalog of Claude Code plugins. One slash command installs GitHub, Slack or Notion as a pre-configured MCP server.
OpenMAIC: A Course Built By An Agent
An agent plans, builds and revises the slides, quiz and interactives. Self-hosted, bring your own model key.
God's Eye View: A Live 3D Globe
Real flights, ships, satellites and public cameras on one photorealistic globe, running in your browser with no API keys required to start.
Modern Go Guidelines: Stop Writing 2019 Go
A JetBrains plugin that teaches Claude Code the Go idioms your model trained before, keyed to your go.mod version.
Scientific Agent Skills: 164 Skills For Real Research
One skill per scientific library or database, not one generic assistant. Install the two your lab actually uses instead of all 164 at once.
MiniMind: Train Your Own Tiny LLM
The full pipeline, pretrain to chatting model, in from-scratch PyTorch. One fine-tuning epoch runs on a single 3090 in about two hours.
OpenSEO: Semrush Without The Subscription
Keyword research, rank tracking, backlinks and audits, self-hosted and billed by usage, not a seat. An MCP server puts it in Claude Code.
Omarchy: Linux Built For Agents
DHH's opinionated Arch Linux with agent tooling built into the desktop. Installs from a bootable USB and can wipe the target drive.
TimesFM: Forecasting Without The Spreadsheet Math
Google's pretrained forecasting model projects any time series forward. The newest weights are non-commercial only, older ones are not.
Claude Code Setup
Blank Mac to production-grade Claude Code install — every config, every alias, every guardrail, in order.
Claude Code Skills: The Complete Guide
Every skill pattern that survives a real codebase — from single-file helpers to multi-agent orchestrations.
Sonnet 4.6 Agent Guide (Claude Code)
What changed in Sonnet 4.6 for agents — context, tools, latency, and the workflow patterns that exploit each one.
Ponytail Repo
Every Claude Code agent, skill, and prompt we ship at scale — a clean, forkable repo. Clone, configure, ship.
Superset.sh: Claude Opus 5 + Codex Agents in Parallel
Run Claude Opus 5 and Codex side-by-side from one shell — task routing, error fallbacks, and the coordination prompts.
Claude Code Agent Swarm
336 blocks on running dozens of Claude Code agents in parallel — without them stepping on each other or your git tree.
Claude Cowork
The two-minute Claude setup that runs your inbox, calendar, and admin on autopilot — drafts pre-written, you just approve.
How to Setup Claude MCP
The first MCP you should ship — install, write a tool, expose it, call it from Claude Code. The rest follows from here.
Docker Desktop MCP
Turn your container stack into Claude tools — bind, expose, call. The MCP that makes Docker a first-class citizen.
Claude Code + n8n MCP Setup Guide
Wire Claude Code straight into n8n — every node, every trigger, every credential pattern that survives a real workflow.
Higgsfield MCP: Complete Setup & Workflow
AI-generated UGC video inside Claude Code — full connector setup, the prompts, and the cost controls that keep it under budget.
GBrain: Self-Wiring AI Memory
Garry Tan's self-wiring memory layer for Claude — the schema, the writes, and the retrieval patterns that make agents feel current.
Self Host n8n Setup Guide
A production n8n install on your own hardware — Docker, TLS, backups, monitoring. Boring, reliable, yours.
The Complete OpenClaw Stack
An open-source alternative to the paid Claude stack — same capability, zero lock-in, fully self-hosted on your own metal.
AI Infrastructure 2026
The hardware, network, and orchestration layer under any serious AI operation in 2026 — sized, priced, and bench-tested.
AI Infrastructure Stack — Complete Architecture Guide
The full architecture in one reference doc: model, vector store, queue, orchestrator, observability — wired and explained.
The Agentic Coding Blueprint
The 250-block playbook for building software with agents — spec, plan, build, review, ship. No magic, just process.
The Agent Playbook
Ninety-one patterns for getting an agent from "kinda works" to "ships to customers" — the playbook I wish I had at number 1.
Stuck on a stack I haven't shipped yet?
I take on a small number of in-person installs and architecture audits each quarter — Claude, n8n, MCP, infra, the awkward bits in between. If you've got a build that needs a senior pair of hands and a Tuesday on-site, send the brief.
Work with me