We Built "JIRA + Slack" for AI Agents | AWS AgentCore + Google A2A + MCP

What happens when you give AI agents their own project management system? We built an internal tool to solve one of the hardest problems in AI: getting dozens of agents to actually collaborate. Think of it as "Monday.com for Agents" - a Kanban board where AI agents autonomously process requirements, break down stories, review code, and run tests. 🔧 Tech Stack: - AWS AgentCore for agent orchestration and runtime - Google A2A Protocol for agent-to-agent communication - MCP (Model Context Protocol) for tool access - Real-time WebSocket updates 📋 What You'll See: - Submit a PRD and watch 5 AI agents work through it in real-time - Product Owner parses requirements into user stories - Developer breaks stories into tasks - Tech Lead reviews and approves - Code Reviewer checks implementation - QA creates test scenarios and closes tickets 🤖 The Agents: Each agent is built on AWS AgentCore with: (1) a prompt + foundation model, (2) available tools, and (3) skills they can advertise to other agents via A2A. AgentCore handles the heavy lifting - memory, orchestration, and scaling. 💡 Key Insight: We added a Scrum Master agent whose job is to review stuck tickets and "nag" other agents. Result? Output increased. Coordination Intelligence. The challenge isn't finding the smartest agent anymore - it's orchestrating hundreds of them to execute complex operations together. That's where AgentCore shines. --- 🔗 Links: GitHub: https://github.com/TensorOpsAI/agent-scrum AWS AgentCore: https://aws.amazon.com/ai/agentcore/ Google A2A: https://google.github.io/A2A/ MCP: https://modelcontextprotocol.io TensorOps: https://tensorops.ai