# About Axis

Axis is a parallel agent orchestration layer: the distributed memory and atomic
task registry that multiple AI coding agents need to work in one codebase without
colliding.

## Why it exists

As AI loops shift from single-agent to parallel execution, coordination becomes
the bottleneck rather than intelligence.

- **Concurrency collision** — without a shared brain, agents overwrite each
  other's work.
- **Task duplication** — agents have no central registry to partition and claim
  specific tickets.
- **Synergy loss** — separate agents cannot share what they have learned in real
  time.

## Capabilities

- **Distributed memory** — a live notepad that synchronises state across agent
  processes, people and machines.
- **Job orchestration** — an atomic, priority-based job board with dependencies.
- **Conflict avoidance** — advisory per-file locks with tamper detection.
- **Context governance** — the project soul: goals and conventions mirrored to
  every agent that connects.

## How it works

1. **Install** — add the hosted MCP server, or run the local stdio server.
2. **Connect** — open the same project in Cursor, Claude Code, Windsurf or any
   MCP client; they share locks and jobs automatically.
3. **Orchestrate** — agents claim tasks from the board, lock files before
   editing, and sync context through the notepad.

Axis exposes its tools over the Model Context Protocol, so it works with any MCP
client. Source is AGPL-3.0 at https://github.com/VirSanghavi/axis. Get started at https://useaxis.dev/docs.