Download
Agent Foundation is a self-hosted platform for building, running, and managing AI agents through a scalable execution environment. Its Service runtime manages agent requests, execution state, and recovery from interruptions. Developers can configure agents and inspect conversations, tool calls, and execution details through a web-based Console. An integrated agent harness provides tool usage, Skills, memory, and subagent capabilities. Long-running tasks can continue after clients disconnect, while durable checkpoints support recovery after worker failures. The platform connects to configurable model, sandbox, web search, and memory providers. Applications interact with agents through an HTTP API or SDKs for Python, TypeScript, Go, and Rust. Deployment options include Docker Compose and Kubernetes.
Features
- Self-hosted AI agent execution with durable checkpoints and automatic failure recovery
- Web-based management Console for configuring, testing, and debugging AI agents
- Integrated agent harness supporting tool calls, Skills, memory, and subagents
- Configurable AI model, sandbox, web search, and memory provider integrations
- HTTP API and SDKs for Python, TypeScript, Go, and Rust application development
- Docker Compose and Kubernetes deployment with execution monitoring and usage tracking
Project Samples
