d:spatch Documentation
Deploy, orchestrate, and monitor autonomous AI agents with d:spatch.
d:spatch is an open-source platform for deploying, orchestrating, and monitoring autonomous AI agents. Agents run in Docker-sandboxed workspaces, communicate via a structured inter-agent protocol, and are managed through a cross-platform app or CLI.
Getting Started
Install d:spatch, create your first workspace, and deploy an agent.
Core Concepts
Workspaces, agent templates, hierarchies, communication, and configuration references.
Agent SDK
Build agents in Python using the d:spatch SDK with Claude, OpenAI, or any LLM.
CLI
Manage agents, workspaces, and sessions from the command line.
Key Features
- Agent Workspaces — Isolated project environments where agents operate on your codebase.
- Hierarchies & Teams — Organize agents into parent-child trees for delegation and coordination.
- Docker Sandbox — Every agent runs inside a Docker container with controlled filesystem access.
- Inter-Agent Communication — A structured protocol for agents to exchange messages, delegate tasks, and share context.
- Inquiry System — Agents can ask the user (or other agents) questions when they need clarification or approval.
- Agent SDK — Python SDK for building agents, with TypeScript and Rust SDKs coming soon.
- CLI — Full command-line interface for managing workspaces, agents, and sessions.
- Logging & Observability — Structured logs, agent state tracking, and real-time output streaming.
- Community Hub — Share and discover agent templates built by the community.
Early Access
d:spatch is in early access (v0.1). Some features like multi-device networking and end-to-end encryption are coming soon.