Getting Started
Overview of CHAOS-AI concepts, architecture, and how to begin.
What is CHAOS-AI?
CHAOS (Context Hydrated Agentic Orchestration System) is a local-first, provider-agnostic multi-agent AI workspace for the terminal. It coordinates 36 specialized agents through a shared, persistent context layer to deliver faster, cheaper, and more accurate AI-assisted development.
CHAOS v4.3.0 ships with 134 MCP tools, 80 skills, 72 slash commands, and 6 provider adapters — all running on your machine.
Core Concepts
Context Hydration
Every agent shares a persistent context layer backed by three databases: a SQLite FTS5 context store, a cross-project Praxis Store, and a PostgreSQL knowledge graph. When one agent discovers something about your project, that knowledge is available to every subsequent agent. This eliminates redundant file reads and cuts token usage by 50%+.
Agent Orchestration
The PM (Project Manager) agent acts as a central dispatcher. It receives high-level tasks, decomposes them into subtasks, and assigns them to specialized agents running as real subprocesses. Agents can run in parallel, with the PM coordinating dependencies and merging results via a PostgreSQL task queue.
MCP Integration
CHAOS exposes its full capability surface via the Model Context Protocol — 134 tools, 19 resources, and 8 workflow prompts. Connect any MCP-compatible editor (Claude Desktop, VS Code Copilot, Cursor, Windsurf) with a single cw mcp command.
Provider Agnosticism
CHAOS works with Claude Code, GitHub Copilot CLI, Gemini CLI, Aider, OpenRouter, and Ollama. Assign different providers to different agents in the same pipeline to optimize for cost, quality, and privacy.
Architecture Overview
| Layer | Purpose |
|---|---|
CLI (cw) | Terminal interface and command routing |
| Engine | PM dispatcher, task queue, event bus |
| Context | SQLite store, Praxis Store, knowledge graph |
| Providers | 6 LLM API adapters |
| Agents | 36 specialized agents |
| MCP Server | 134 tools via FastMCP |
| Web UI | Dashboard, analytics, event stream |
Next Steps
- Installation — Install CHAOS-AI on your system
- Quick Start — Up and running in 5 minutes
- Configuration — Full configuration reference
- Agents — The full agent catalog
- MCP — Connect your editor via MCP