I build AI tooling for developers and full-stack applications for clients. Away from the keyboard, I design precision telescopes, 3D print them, and point them at things that are very far away.
I got into engineering through marketing, found myself more interested in building the tools than using them, and never looked back.
These days I'm wearing a few hats: Head of Engineering at Hoopless, co-founder at Hoopless.ai, and Head of Engineering at Sky High Visions, where I manage a small team delivering across 30+ client engagements while building full-stack applications, Shopify apps, and custom automations as an IC. Previously I was at Loop, building post-purchase tooling for 5,000+ DTC brands, and WillowTree, doing platform engineering for Fortune 500 clients.
Outside of work I maintain a suite of Go daemons that pre-compute context for AI coding agents, a web scraper built for LLM pipelines, and a couple of 3D-printed telescope designs on MakerWorld.
$ scry status › repos indexed 4 › symbols cached 148,291 › avg query 2.1ms › MCP server running $ █
One Go daemon spanning five domains: code symbols, git history, database schemas, HTTP traffic, and a cross-domain graph that connects them. 23 MCP tools serving pre-computed answers in single-digit milliseconds. Replaces 30-50% of an AI agent's tool calls, and replaced three of my own earlier daemons.
Repository →$ trawl --help › intelligent tiered web scraping › › scrape Scrape a single URL › crawl BFS-crawl a site › batch Scrape URLs from a file › map Enumerate every URL on a site › sitemap Discover URLs from sitemaps › mcp Run as MCP server › resume Resume interrupted job $ █
A single Go binary for tiered web scraping, BFS crawling, schema extraction, and clean markdown. Polite by default with opt-in evasion tiers, per-host tier learning, persistent job frontier, and structured JSONL output ready for LLM pipelines. No API key, no recurring cost.
Repository →There are hundreds of MCP servers now. Most aren't worth your time. These are the ones I actually use, have built, or would recommend to another developer who ships code daily.
8 minGit data is rich but unstructured. Lore is an MCP server that pre-indexes blame, history, and co-change patterns into BadgerDB and serves them as seven focused tools, giving agents structured git intelligence instead of raw CLI output.
8 minA practical walkthrough for configuring MCP servers in Claude Desktop and Claude Code: where the config lives, what the JSON looks like, and how to debug when things don't connect.
8 minThe protocol layer of an MCP server is trivial. The hard part is designing tool interfaces that agents actually use well: narrow tools, precise descriptions, structured returns, and sub-10ms latency that changes agent behavior.
11 minMost MCP servers run on a developer's laptop. When you need a team to share one, everything changes: transport, auth, lifecycle, and monitoring all become real problems.
8 minEvery image here was shot through a telescope I designed in Fusion 360 and printed in PET-CF. Most are LRGB or narrowband composites stacked from 4-14 hours of 300-second subs.
The more concrete your note, the faster I can help. I answer everything thoughtfully within a few days.
jeff@hooton.codes →