How do you build the engineering system around AI coding agents for production?
Key_Ad_2747 · reddit · 2026-09-11
A web developer who uses AI coding agents daily argues the hard part is no longer getting agents to write code, but building the system around them. His setup includes per-repo knowledge bases, reusable skills/rules encoding conventions, automated local onboarding, a structured issue → dev → verification → completion workflow, and mandatory pre-completion checks — with the goal of making the agent itself replaceable.
He admits the setup is still cumbersome: onboarding isn't deterministic, context grows too much, rules overlap, and manual intervention remains heavy.
He's also looking for a real control plane above GitHub and coding agents: manage multiple repos, dispatch tasks to different agents in parallel, isolate each task in its own branch/worktree, review progress and PRs, keep GitHub Issues as the source of truth, and stay agent-agnostic. Does something like this exist in practice, or is everyone building their own orchestration layer?
More from coding & agent
- 'Let an agent run wild on vague goals and it makes your code worse 9 out of 10 times' — eptwts · 2026-10-27
- LangChain to demo the in-house design agent its team uses daily, this Wednesday — LangChain · 2026-10-27
- Biotech Agent Series Day 8: A Notebook Review Agent to Auto-Audit Lab Records — nlarusstone · 2026-10-27
- Photon Launches Agent-to-Agent Messaging in iMessage for Real AI Collaboration — alvelda · 2026-10-27
- Fake-paper meme shows an agent workflow for reverse engineering trading systems into a one-person quant fund — PtrPomorski · 2026-10-27
- Codex clones Linear's core from a one-page spec in 3 minutes 28 seconds — aakashgupta · 2026-10-27