A 31-rule prompt that runs your coding agent as an accountable engineering owner
boringmarketer · x · 2026-09-05
boringmarketer shares what he calls the best coding prompt for GPT-6 Astra: a 31-rule instruction that treats the AI as the accountable engineering owner of a repository.
Key ideas:
- Inspect before changing: study the product, architecture, agent instructions, skills, tests, and CI; find what causes unfinished work and repeated confusion; ship small reviewable improvements first.
- Consolidate over adding: prefer clarifying existing mechanisms; add new process only for demonstrated problems; avoid duplicated sources of truth; keep main agent instructions short and link out deeper guidance.
- Define completion by observable user outcomes: trace changes through interface, backend, persistence, workers, and external services; state what was exercised and what remains unverified.
- Delegate in bounded chunks only when it reduces total effort; avoid recursive delegation and review ping-pong.
- Match verification to risk, separate blocking defects from optional improvements, never merge or deploy without authorization, and keep regression tests unless proven obsolete.
His advice to start: pick one real feature and measure how much human intervention it still needs.
More from coding & agent
- GPT-6 Astra builds a procedural three.js game under 2MB, then remakes it in Blender — gabrielchua · 2026-09-05
- OpenAI brings ChatGPT and Codex into UPenn drug discovery workflows — AllThingsApx · 2026-09-05
- Codex /fast mode runs at 2x speed and stays within subscription, unlike Claude's paid extra usage — rudrank · 2026-09-05
- ChatGPT Workspace 5-hour limit mid-agent-run wipes uncommitted code, dev demands fix — cneuralnetwork · 2026-09-05
- Astra costs ~$1.49/M tokens for this dev's coding — but is it good enough to be the default? — notchillatall · 2026-09-05
- HeroUI launches Agents: embed a context-aware agent in your product — ycombinator · 2026-09-05