Articles for people who
ship the stack
Rewrites on React, Node.js, TypeScript and AI. A translated body is shown when it exists. If it does not, you see the English text and a note that says so.
Tagged: AI
Coverage Is Not a Decision: Put the Pick on the Page
A complete comparison table isn't a recommendation. Why AI-made option menus push editing work downstream, and why the pick still matters on the page.
5 min readWhy Compiler Theory Still Matters When AI Writes the Code For You
AI can generate a parser in minutes, but it won't tell you when you need one. See how small languages and deterministic rules beat prompts, and why you're the bottleneck.
5 min readPrompt Engineering for DevOps: The C.L.E.A.R. Checklist and a Prompt Library
Practical prompt engineering for DevOps: a five-part checklist, before and after examples for Terraform, Kubernetes and CI, plus tips for building a reusable prompt library.
5 min readJev by TypeSafe: A Model That Returns Typed Decisions Instead of Text
Jev drops text generation and returns typed choices, scores and calibrated probabilities in one fast call. See how the API works, what testers found and where it falls short.
6 min readShip Agent Plugins, Not Naked MCP Servers
Ship MCP tools as agent plugins with skills, permissions and install metadata, so hosts get a finished capability instead of a bare MCP server adapter.
11 min readWho writes the code when agents join the workflow?
From autocomplete to agent swarms: map each coding agent tier to blast radius, test coverage and review gates, so humans stay on the code that really matters.
10 min readSkills plus MCP: tools give hands, skills teach the workflow
MCP gives agents tools; skills teach the workflow: which tools to call, in what order, and when to stop for approval. Why skills plus MCP beats either one.
4 min readCompress every LLM call at the LiteLLM gateway with Headroom
Headroom as a LiteLLM pre-call guardrail compresses every LLM call: reversible compression, fail-open design, per-key rollout and the token savings measured.
7 min readIntroduction to MCP: from fragmentation to shared tool servers
Why pasting context fails, how function calling helped, why N×M integrations hurt, and how MCP moves an LLM stack toward N+M reusable shared tool servers.
6 min read
About these articles
Request a 24h estimate
Need the same stack in a production operator layer? Send the brief — estimate within 24 hours.