# Ash

> Ash is an open-source CLI, SDK, and self-hostable system for deploying and orchestrating AI agents. Deploy Claude agents as production APIs with sessions, streaming, sandboxing, and persistence.

## Getting Started

- [Introduction](https://docs.ash-cloud.ai/): Overview of Ash and what it does
- [Installation](https://docs.ash-cloud.ai/getting-started/installation): Install the Ash CLI and SDK
- [Quickstart](https://docs.ash-cloud.ai/getting-started/quickstart): Deploy your first agent in minutes
- [Key Concepts](https://docs.ash-cloud.ai/getting-started/concepts): Agents, sessions, sandboxes, and streaming
- [Use Cases](https://docs.ash-cloud.ai/use-cases): Common patterns and examples for building with Ash

## Guides

- [Defining an Agent](https://docs.ash-cloud.ai/guides/defining-an-agent): Create agent folders with CLAUDE.md and config
- [Deploying Agents](https://docs.ash-cloud.ai/guides/deploying-agents): Deploy agents to an Ash server
- [Managing Sessions](https://docs.ash-cloud.ai/guides/managing-sessions): Create, resume, and end agent sessions
- [Streaming Responses](https://docs.ash-cloud.ai/guides/streaming-responses): Real-time SSE streaming from agents
- [Working with Files](https://docs.ash-cloud.ai/guides/working-with-files): Upload and download files from agent sandboxes
- [Authentication](https://docs.ash-cloud.ai/guides/authentication): API keys and multi-tenant auth
- [Monitoring](https://docs.ash-cloud.ai/guides/monitoring): Health checks, Prometheus metrics, and structured logs

## Self-Hosting

- [Docker (Default)](https://docs.ash-cloud.ai/self-hosting/docker): Run Ash with Docker sandbox isolation
- [Deploy to AWS EC2](https://docs.ash-cloud.ai/self-hosting/ec2): Production deployment on EC2
- [Deploy to Google Cloud](https://docs.ash-cloud.ai/self-hosting/gce): Production deployment on GCE
- [Configuration Reference](https://docs.ash-cloud.ai/self-hosting/configuration): All environment variables and settings
- [Multi-Machine Setup](https://docs.ash-cloud.ai/self-hosting/multi-machine): Scale with separate server and runner nodes

## API Reference

- [API Overview](https://docs.ash-cloud.ai/api/overview): REST API conventions and base URL
- [Agents](https://docs.ash-cloud.ai/api/agents): List and inspect deployed agents
- [Sessions](https://docs.ash-cloud.ai/api/sessions): Create, get, list, and delete sessions
- [Messages](https://docs.ash-cloud.ai/api/messages): Send messages and stream responses via SSE
- [Files](https://docs.ash-cloud.ai/api/files): Upload and download sandbox files
- [Health and Metrics](https://docs.ash-cloud.ai/api/health): Health check and Prometheus metrics endpoints

## SDKs

- [TypeScript SDK](https://docs.ash-cloud.ai/sdks/typescript): Official TypeScript/Node.js client
- [Python SDK](https://docs.ash-cloud.ai/sdks/python): Official Python client
- [Direct API (curl)](https://docs.ash-cloud.ai/sdks/curl): Use the REST API directly with curl

## CLI Reference

- [CLI Overview](https://docs.ash-cloud.ai/cli/overview): Command-line interface reference
- [Server Lifecycle](https://docs.ash-cloud.ai/cli/lifecycle): Start, stop, and manage the Ash server
- [Agent Commands](https://docs.ash-cloud.ai/cli/agents): Deploy, list, and remove agents
- [Session Commands](https://docs.ash-cloud.ai/cli/sessions): Manage sessions from the CLI
- [Health](https://docs.ash-cloud.ai/cli/health): Check server health from the CLI

## Architecture

- [System Overview](https://docs.ash-cloud.ai/architecture/overview): How Ash components fit together
- [Sandbox Isolation](https://docs.ash-cloud.ai/architecture/sandbox-isolation): Docker and bubblewrap isolation model
- [Bridge Protocol](https://docs.ash-cloud.ai/architecture/bridge-protocol): Unix socket protocol between server and sandbox
- [Session Lifecycle](https://docs.ash-cloud.ai/architecture/session-lifecycle): State machine for session management
- [Sandbox Pool](https://docs.ash-cloud.ai/architecture/sandbox-pool): Pre-warming and pool management
- [SSE Backpressure](https://docs.ash-cloud.ai/architecture/sse-backpressure): Flow control for streaming responses
- [Database](https://docs.ash-cloud.ai/architecture/database): SQLite schema and persistence
- [Scaling Architecture](https://docs.ash-cloud.ai/architecture/scaling): Multi-machine scaling design
- [Design Decisions](https://docs.ash-cloud.ai/architecture/decisions): ADRs and architectural rationale

## Optional

- [Ash vs ComputeSDK](https://docs.ash-cloud.ai/comparisons/computesdk): Feature comparison with Anthropic ComputeSDK
- [Ash vs Blaxel](https://docs.ash-cloud.ai/comparisons/blaxel): Feature comparison with Blaxel managed agent platform
- [Development Setup](https://docs.ash-cloud.ai/contributing/development-setup): Set up the monorepo for local development
- [Project Structure](https://docs.ash-cloud.ai/contributing/project-structure): Package layout and responsibilities
- [Testing Guide](https://docs.ash-cloud.ai/contributing/testing): How to write and run tests
- [Release Process](https://docs.ash-cloud.ai/contributing/releases): Changesets, versioning, and publishing