A Tiny HTTP Server in Yar
Yar can now serve HTTP. Not a framework, not a router, not a production web stack. Just a small stdlib package that wraps TCP, parses one request, calls a handler, writes a response, and closes the connection.
I write about backend systems, coding agents, infrastructure, CLI tools, and the boring details that decide whether software works in production.
Latest writing
-
-
AI Coding Agents Need a Source of Truth
Bigger prompts do not fix drifting agent plans. A good workflow starts with a small brief, human review, task sizing, and checks against real artifacts.
-
Good CLI Design Is Mostly Silence
· cli
A CLI is both a user interface and an API for scripts. Quiet mode, stdout, stderr, colors, and exit codes need real contracts.
-
The Build Cache Was Not Broken
Docker cache problems are often Dockerfile problems: unstable inputs, bad COPY order, and layers that depend on too much.
-
How to Cook Claude Code
· llm
Claude Opus 4.7 scores 87.6% on SWE-Bench Verified. Your daily experience probably does not. The gap is not only the model. It is the setup around it.
-
It's Not the Model. It's the System Prompt You Never Wrote.
· llm
Most complaints about coding agents are complaints about empty context. CLAUDE.md, distilled project docs, and a few slash commands change the same model from autocomplete into something useful.
-
I Built an FPS in Godot. My Daughter Is Building the World.
A backend engineer with no game dev background picked up Godot 4 and built a horror FPS. Then a 13-year-old played it, grabbed a tablet, and decided the game needed real monsters and lore.
-
Preview Environments Are Easy to Want and Hard to Operate
The pitch is obvious: per-PR environments, live links, faster feedback. Then a multi-service stack needs twenty restarts and the pitch meets operations.
-
The CI Pipeline Nobody Owns
Your CI pipeline builds the same Docker image three times, a dependency bot eats your runner fleet, and nobody fixes it because nobody owns the system.
-
We Added Monitoring to Everything (Except the Thing That Kept Breaking)
Production has dashboards, alerts, and logs. Test environments have SSH. CI runners have a re-run button. That gap quietly eats engineering time.