What I learned building an opinionated and minimal coding agent
mariozechner.at · 7,008 words · saved by 1 readers
Lessons I learned while building my own coding agent from scratch.
What I learned building an opinionated and minimal coding agent 2025-11-30 It's not much, but it's mine Table of contents pi-ai and pi-agent-core There. Are. Four. Ligh... APIs Context handoff We live in a multi-model world Structured split tool results Minimal agent scaffold pi-tui Two kinds of TUIs Retained mode UI Differential rendering pi-coding-agent Minimal system prompt Minimal toolset YOLO by default No built-in to-dos No plan mode No MCP support No background bash No sub-agents Benchmarks In summary In the past three years, I've been using LLMs for assisted coding. If you read this, y
saved by
related reading
- Shipping at Inference-Speed | Peter Steinbergersteipete.me
- A Guide to Claude Code 2.0 and getting better at using coding agents – sankalp's blogsankalp.bearblog.dev
- How I Use Claude Code | Philipp Spiessspiess.dev
- Building Effective AI Agents \ Anthropicanthropic.com
- Best practices for Claude Code - Claude Code Docsanthropic.com
- Don’t Outsource Your Thinkingteltam.github.io
- Building Effective AI Agents \ Anthropicanthropic.com
- Ampampcode.com
- Designing agentic loopssimonwillison.net
- Here’s how I use LLMs to help me write codesimonwillison.net
- Towards self-driving codebases · Cursorcursor.com
- How I'm using coding agents in September, 2025 — Massively Parallel Procrastinationblog.fsck.com