The Mega Repo Manifesto 🚀

Welcome to the post-polyrepo era. Mega Repo is a coding context system for AI agents.

🧠 The Problem: Conway's Law is Blinding Your AI

We split codebases into microservices and polyrepos because human working memory is limited. We use physical isolation to reduce cognitive load. Even in a monorepo, critical engineering context still lives outside source code. AI models in 2026 can handle 1M+ context windows. They excel at cross-domain reasoning. When you artificially fragment your codebase, you are blindfolding your AI agent. An AI harness without global context is just a brain in a jar.

🌍 The Solution: Mega Repo

To give an AI agent "God Mode," it needs the complete world state. A Mega Repo puts everything into one physical workspace: frontend, backend, AWS configs, Kubernetes manifests, test automation, and context from Jira, Linear, or Slack.

⚡ 0-Cost Setup (The Hack)

You don't need to refactor your company's infrastructure. Build a Mega Repo locally:

1. Clone this repository.
2. git clone your work repositories into sources/ (or use /repo-add <git-url> [target-name]).
3. Open a terminal at the root of this Mega Repo, then launch claude or codex.

Read AGENTS.md to see how your AI should operate in this environment.

🛠 Skills & Extensibility

This repo uses Agent Skills. If your AI needs to query specific PostHog dashboards or sync with Jira, it can learn by doing, or you can add related skills directly.