An Ex-Meta L8’s Agentic Engineering Setup

2025-12-15 · Source: ByteByteGo Newsletter · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Software Development & Engineering · Depth: Advanced, extended

Summary

Kun Chen, a former L8 principal engineer at Meta, Microsoft, and Atlassian, details his agentic engineering setup that enables him to ship over 30 high-quality pull requests on a "slow day". His workflow is built around a terminal-centric environment, utilizing WezTerm, Neovim with plugins like oil.nvim and neogit, and Tmux for robust session management. Key to his productivity is voice input via OpenSuperWhisper for prompting agents, treating them like a team to whom he delegates outcomes rather than actions. For complex projects, he employs his open-source Lavish Editor for interactive, visual planning with agents. Autonomous implementation is handled by `gnhf` for large tasks, while `no-mistakes` performs peer reviews, end-to-end testing, and PR generation, identifying bugs in 68% of changes. Parallel work is facilitated by Tmux and `treehouse` for managing Git worktrees, and remote access is maintained through Tailscale and mosh for seamless continuity across devices.

Key takeaway

For software engineers aiming to significantly scale their development output with AI agents, you should adopt a manager-like delegation approach, focusing on defining outcomes rather than micro-managing actions. Implement a robust, terminal-centric setup with tools like Tmux and Git worktrees to parallelize tasks. Critically, integrate automated planning and validation pipelines, such as Lavish Editor for interactive design and `no-mistakes` for E2E testing and peer review, to ensure quality and autonomy. This shift allows you to manage multiple projects concurrently, minimizing context switching and maximizing agent utility.

Key insights

High-productivity agentic engineering stems from manager-like delegation to AI agents, supported by specialized, agent-agnostic tooling for planning, execution, and validation.

Principles

• Delegate outcomes, not specific actions, to AI agents.
• Maintain agent-agnostic tooling to ensure model flexibility.
• Automate validation with E2E tests and peer-review agents.

Method

The workflow integrates voice input for prompting, interactive planning via HTML-based tools, autonomous implementation with orchestrators, and automated E2E validation and PR generation using specialized review agents.

In practice

• Use OpenSuperWhisper for voice-driven prompting.
• Leverage Tmux and Git worktrees for parallel development.
• Implement `no-mistakes` for automated code review and testing.

Topics

• Agentic Engineering
• Developer Productivity
• AI Agents
• Terminal Workflow
• Automated Code Review
• Git Worktrees

Code references

• stevearc/oil.nvim
• neogitorg/neogit
• folke/snacks.nvim
• starmel/OpenSuperWhisper

Best for: AI Engineer, Software Engineer, MLOps Engineer

Related on AIssential

• Fragments: April 29 — Agentic engineering shifts focus from building fast to verifying fast, emphasizing automated checks and clear definitions.
• The Semi-Executable Stack: Agentic Software Engineering and the Expanding Scope of SE — Agentic AI expands software engineering's scope to "semi-executable artifacts" beyond traditional code.
• The AI-First Data Engineer: 10–50x Productivity and What Changes Next — Agentic AI transforms data engineering by automating entire workflows, shifting human roles to agent operation and increasing data product creation.
• You NEED to know these vibe coding secrets — Structured agent configuration, reusable skills, and continuous loops automate advanced AI coding workflows.
• Agentic AI That Actually Works — Enterprise AI success requires embedding agentic AI into governed workflows for controlled, scalable, and auditable execution.
• What Developers Actually Need to Know Right Now — AI's true frontier lies in orchestrating agents to solve real problems with control and traceability.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by ByteByteGo Newsletter.