AgentOrch is a production-ready AI Agent Orchestration Platform built with .NET 10 and the Model Context Protocol (MCP). It enables teams to coordinate fleets of specialized AI agents — Orchestrator, CodeGenerator, TestGenerator, Reviewer, Deployer, and Monitor — each communicating over a standard MCP protocol. Tasks are structured hierarchically (main task with subtasks), can be paused mid-execution with dynamic instruction updates, and resumed seamlessly. Real-time visibility is provided via Server-Sent Events (SSE), streaming live agent conversations and activity logs to a built-in dashboard. BDD/Gherkin support lets teams define acceptance criteria before execution, closing the loop between specification and implementation. A complete audit trail and event-driven architecture make the platform suitable for regulated environments and continuous-delivery pipelines.
Visit Project ↗