AI Skills
The AI skills I have built and use every day to ship across the portfolio. Some are knowledge skills that advise. Some are agents that do the work. This is what each one does.
Most are agents that do the work; a few are knowledge skills that advise.
Research and review skills triage the backlog and red-team the plan.
Build and dispatch skills run the fleet of agents in isolation.
Content, SEO, and design skills build and publish the work.
Analytics and memory skills check what shipped and remember it.
one loop · repeats continuously
The skills that keep every page strong for search and answer engines.
Grades a page or a pre-publish draft for SEO, AEO, and GEO, then lists the cheapest high-leverage upgrades. Blocks content that would hurt the site.
Assembles one site's SEO, answer-engine, and AI-visibility scorecard with a prioritized, leverage-ranked fix plan.
Finds thin or near-duplicate pages across a site and recommends what to index, noindex, or merge.
Audits a page against modern search ranking factors and explains why it ranks or does not.
Turns a full site crawl into tracked, dispatchable fixes, routed per page.
Checks whether answer engines cite a page and what to change so they do.
How one operator runs a fleet of coding agents without collisions.
Turns a goal into paste-ready prompts for five to ten autonomous coding agents, each right-sized to a model and isolated in its own workspace.
Checks whether a piece of work is already in flight before I start, so parallel sessions never duplicate each other.
Measures a real coverage baseline, then writes tests module by module, with a mutation check that rejects vacuous tests.
Gives every file-writing agent its own workspace so parallel work never corrupts a shared checkout.
Skills that pressure-test a plan and keep the backlog honest.
Red-teams a plan before I commit, grounded in my own catalogued past mistakes, then returns a ranked risk list and a go or no-go call.
Audits my notes and docs for stale, contradictory, or duplicated entries and proposes a prune.
Scans a full initiative or issue backlog and proposes a ranked, machine-readable action plan. Proposes only.
Runs multi-source research and synthesizes a cited, narrative report.
Skills that remember work across sessions and machines.
Captures what a work session shipped and files it to a searchable timeline I can read on any device.
Finds and reloads a past work thread with its exact next action, so nothing restarts from memory.
Returns a short, grounded status on any topic, pulled from durable signals rather than guesswork.
Routes a question to a phone-answerable queue and syncs the answer back to the agent that asked.
Skills that keep numbers canonical and decisions logged.
Pulls a single project's traffic and search data and returns only that project's slice.
Reads canonical portfolio counts so figures in content are never hand-typed or invented.
Spawns a click-through form for a set of grouped decisions and persists the answers for history.
Skills that make the output read as human and map how a product works.
Edits a draft to read as human, sharp, and opinionated while keeping the writer's own voice.
Scans a codebase and writes a self-contained service-blueprint journey map.
I build production Django / Next.js platforms and human-supervised AI-agent systems. Solo, senior, and fast. Tell me what you are building.