AgentSystems: The Open-Source Runtime for Self-Hosted AI Agents

AgentSystems is an open-source, self-hosted application platform and runtime designed for third-party AI agents, offering a decentralized and transparent infrastructure for building, deploying, and managing AI agents without relying on centralized gatekeepers. Built for developers and technical teams, it supports a wide range of deployment environments including macOS, Linux, cloud platforms (AWS, Azure, GCP), and on-premises or air-gapped networks, enabling secure and scalable agent execution. The platform abstracts AI provider integrations (OpenAI, Anthropic, AWS Bedrock, Ollama) and provides robust features like isolated Docker containers, runtime credential injection, configurable egress controls, and hash-chained audit logs for accountability. It also includes a comprehensive suite of tools including a FastAPI gateway, CLI, SDK, web UI, Python toolkit, and a Git-based agent index for publishing and discovering agents. AgentSystems is ideal for developers using frameworks like LangChain and LangGraph who want to deploy agents with full control over infrastructure, security, and data flow.

Key Features:

• Self-Hosted Runtime: Deploy agents on your own infrastructure, including servers, VMs, private clouds, and air-gapped networks.
• Open-Source & Apache-2.0 Licensed: Free to use, modify, and distribute with full transparency.
• Agent Index & Publishing: Publish agents via GitHub with automated validation and browse available agents at agentsystems.github.io/agent-index.
• Isolated Execution: Agents run in secure, isolated Docker containers with configurable egress filtering and thread-scoped artifact storage.
• Runtime Credential Management: Credentials are injected at runtime, ensuring user control and security.
• Egress Controls: Configurable default-deny allowlists per agent for network security.
• AI Provider Abstraction: Supports multiple AI providers (OpenAI, Anthropic, AWS Bedrock, Ollama) with inherited user configurations.
• Audit Logging: Hash-chained audit logs (SHA-256 in PostgreSQL) for transparency and accountability.
• Standardized Deployment: Install with a single command: curl -fsSL https://github.com/agentsystems/agentsystems/releases/latest/download/install.sh | sh.
• Comprehensive Tooling: Includes CLI, SDK, web UI (React dashboard), FastAPI gateway, Python toolkit, agent template, and documentation.
• Framework Compatibility: Supports building agents with LangChain, LangGraph, and FastAPI.
• Community-Driven: Active development with support via GitHub, Discord, and documentation.

Pricing: AgentSystems is free and open-source under the Apache-2.0 license, with no paid tiers. It offers a freemium model through its open-source nature and community support, allowing users to self-host and extend the platform without cost.

Conclusion: AgentSystems is a powerful and flexible open-source platform that empowers developers to build and deploy AI agents with full control, security, and transparency. By eliminating gatekeepers and providing a standardized, self-hosted runtime, it represents a critical step toward a commoditized and decentralized future for AI agents.