An open-source TypeScript-first platform for building, testing, running, and orchestrating AI agents. From no-code desktop apps to production-ready servers, TsAgent provides a complete ecosystem for intelligent agent development.
TypeScript agent framework - the core library for building, testing, and running AI agents programmatically.
Desktop application for creating, testing, and managing no-code agents through a visual interface.
Command-line interface for agent operations, automation, and integration with development workflows.
Expose agents as HTTP endpoints using the A2A protocol for production deployment and integration with external systems.
MCP server for orchestrating multiple A2A agent servers, enabling complex agent workflows and multi-agent systems.
References provide persistent memory while Rules enable dynamic learning and behavior guidance through configurable system prompts.
Full MCP protocol support with extensible architecture for real-world interaction, security controls, and container support.
Support for Anthropic, OpenAI, Google, AWS Bedrock, and Ollama with flexible model selection and cost optimization.
Chain agents together using the A2A protocol for complex workflows, multi-agent systems, and intelligent automation.
Developers and researchers building intelligent AI agents need a comprehensive platform for rapid prototyping, memory management, learning systems, and tool integration.
Researchers working with AI models need sophisticated interaction capabilities for data analysis, literature review, experiment management, and knowledge synthesis.
Organizations automating complex business processes including document processing, customer support, data processing, and decision support.
Educational institutions and training programs for AI education, research projects, custom training, and collaborative learning.
Watch an 8-minute demo showing how to build intelligent agents with TsAgent:
Note: In this video the product TeamSpark Workbench is now called TsAgent Foundry and a Workspace is now an Agent.
The central workhorse where you interact with AI models, explore ideas, and solve problems
How your agent collects and remembers information, serving as the "memory" of a workspace
Govern how your agent instructs the AI to process instructions and commands
Specialized functions that extend the AI's abilities beyond just conversation
Support for a wide range of AI model providers giving you flexibility to choose the best model
Use TsAgent Foundry desktop app or CLI for no-code agent creation
Write prompts, engineer context (rules/references), and add relevant MCP tools
Chat with your agent to refine its behavior and capabilities
Chain agents together using the A2A protocol for complex workflows
Embed agents in applications via @tsagent/core API or expose as A2A servers
Get the desktop application for your operating system and start building intelligent agents today.