Introducing Automatic: One Config for Every Agent

Managing AI coding agents in 2026 shouldn't feel like juggling config files across a dozen tools. But that's exactly what it became.

The Problem

As Claude Code, Codex, Cursor, and other AI coding agents landed in quick succession, the promise was incredible: AI that could write, review, and ship code alongside you. The reality was a maintenance tax no one signed up for.

Every agent had its own config format. Every workspace needed its own set of instructions. Your MCP servers had to be registered four separate times. When you found a great Skill, you copied it by hand into each tool.

It was tedious, error-prone, and it scaled badly.

What We Built

Automatic is a single source of truth for your AI agent configuration.

• Skills — curated, reusable prompting libraries that extend what your agents can do
• MCP Servers — register once, sync everywhere
• Instructions — write your agent guidance in one place and push it to every tool automatically

The moment you connect Automatic, your agents are in sync. Add a Skill from the store, and Claude Code, Cursor, and Codex all get it immediately.

How It Works

Automatic runs as a lightweight background daemon. It watches your config and propagates changes to each connected agent in real time. There's no polling, no manual export step, no stale configs.

Under the hood we use the Model Context Protocol (MCP) to talk to agents that support it, and native config file formats for the rest. The abstraction is invisible — you just manage one config.

What's Next

We're shipping the Skill Store alongside the initial release — a growing catalogue of community-built Skills you can install in one click. Next up: team sharing, org-level config, and a CLI for power users.

We're just getting started. If you're drowning in agent configs, download Automatic and let us know what you think.