Developer shares agent system design principles: defaults first, prompts as the customization layer
menhguin · x · 2026-09-21
Author menhguin shares design philosophy from the ehoy project:
- First-party defaults wherever possible: use default settings and commands unless explicit customization is intended, so agents can navigate based on post-training "instincts" and avoid maintaining custom scripts and documentation. The author keeps almost no custom settings on his Mac.
- Prompts as the preferred customization layer: heavy customization lives in prompts because they're self-documenting, portable, and create no maintenance debt.
- Prefer official/well-known defaults over custom solutions when choosing providers or tools.
- Context-rich descriptions explaining why are more durable; each process principle exists because a specific failure taught it.
Excerpted from the author's extremely long system prompt — a hands-on summary of agent engineering workflow practice.
Related event: Engineering Philosophy for the Agent Era: Defaults First(2 posts)→
More from coding & agent
- Turn any local LLM into a confidence-scored classifier via logprobs, full llama.cpp recipe included — DivideHorror3217 · 2026-09-21
- Ruff author charliermarsh: he only started using agents meaningfully in December 2025 — charliermarsh · 2026-09-21
- Open-source LLaMA-Factory fine-tunes 100+ LLMs; 200 examples can beat frontier models — Roger_M_Taylor · 2026-09-21
- SWE-2 free across Devin Cloud Agents, CLI and Desktop until October 8 — silasalberti · 2026-09-21
- Delta and OpenTable block AI agents, hinting at a coming platform stand-off — Scobleizer · 2026-09-21
- ForgePoint Signal launches MCP-native daily US estate and tax law monitoring — modelcontextprotocol · 2026-09-21