Hey PH! I'm Avi — I ship production code on Valkey/GLIDE + AWS by day and live inside AI coding tools by night. I built AgentSys because AI can write code now, but everything *around* the code — picking the task, managing branches, multi-agent review, CI failures, PR comments, cleanup — still requires constant babysitting. AgentSys is the runtime that handles all of it. You type one task. It autonomously: explores your codebase → plans → implements → runs multi-agent review → fixes CI → ships the PR. Zero manual steps in between. Under the hood: 14 plugins, 43 agents, 30 skills — each with a single responsibility. Install via Claude Code's plugin marketplace or npm: • Claude Code: /plugin marketplace add agent-sh/agentsys • All platforms: npm install -g agentsys (then interactive installer for Claude Code, OpenCode, Codex CLI, Cursor, or Kiro) Built on agnix-validated configs so nothing silently breaks. Tested on 1,000+ repos. 3,750 tests passing. Happy to answer anything — which AI coding tool are you using? 👇

AgentSys

AI writes code. AgentSys automates everything else.

AI writes code. AgentSys automates everything else.