Claude Skills
Reusable bundles of instructions, tools, and context that turn Claude into a domain expert. I write and ship them daily — the AI Workflow in a Week tier ships you one.
// concepts
Most AI consultants will sell you a workflow. I'll show you the whole graph it lives inside — products, people, skills, edges, the way the work connects. This is mine. Yours can look like this too.
// 3d · the concepts I build with
// concept catalog
Reusable bundles of instructions, tools, and context that turn Claude into a domain expert. I write and ship them daily — the AI Workflow in a Week tier ships you one.
Multi-agent orchestrators with shared filesystem memory. Used by Anthropic for agents that survive across sessions. I build these for clients who need work to compound.
Model Context Protocol — how Claude talks to external systems (Stripe, Gmail, your database). I write MCP servers and integrate existing ones.
Force-directed graphs of products, people, skills, and edges. My own one (Cortex) has 232 nodes and 1,489 edges. Powers the agents that run my businesses.
Structured data that gets you cited by ChatGPT, Perplexity, and Claude. Schema layering yields ~4.2x higher AI citation rate. Shipped on this site.
Test your agent against a benchmark before shipping. Same logic as TDD, applied to AI behavior. Cuts hallucination, accelerates iteration.
Claude controlling apps on your machine — Chrome, Mail, Slack, native apps. I build automation flows that run while you sleep.
232-node knowledge graph at ~/cortex/. Tracks every product, person, skill, and revenue loop. Auto-heals weekly. The brain that runs the practice.