AMOS: An Automated Development Workflow Orchestration System Based on Multi-Agent Collaboration

AMOS (Autonomous Multi-Agent Orchestration System) is an innovative multi-agent orchestration platform. Through the collaboration of three core agents—Planner, Executor, and Verifier—it converts natural language instructions into complete development workflows, enabling automated integration of tools like GitHub, Jira, and Slack. This addresses the pain point of developers frequently switching between tools and improves development efficiency.

In modern software development teams, developers need to frequently switch between multiple tools every day: managing code repositories on GitHub, tracking task progress in Jira, and communicating with the team on Slack. These repetitive coordination tasks not only consume a lot of time but also easily lead to information desynchronization due to human oversight. AMOS emerged to simplify multi-tool operations through multi-agent collaboration.

AMOS adopts a three-tier agent architecture:

• Planner Agent: Analyzes user natural language instructions and breaks them down into structured execution steps (e.g., deploy new features → update tasks → notify the team).
• Executor Agent: Calls APIs of GitHub/Jira/Slack and other tools to execute plans and collect results.
• Verifier Agent: Validates execution results, triggers retries or alerts to ensure reliability.

• Backend: Spring Boot framework, ensuring stability and concurrent processing capabilities.
• Frontend: Dashboard built with React + TailwindCSS, displaying instructions, execution flow, and status scores.
• AI Model: Mistral 7B is selected, balancing capability and deployment cost, supporting complex task reasoning.

Scenario 1: One-Click Deployment and Notification

When a user inputs 'Deploy the new feature and notify the team', the system executes:

1. The Planner generates a three-step plan;
2. The Executor triggers GitHub deployment, updates Jira tasks, and sends Slack notifications;
3. The Verifier validates the results of each link.

Scenario 2: Cross-Tool Status Synchronization

After a PR is merged, automatically update Jira tickets, notify Slack stakeholders, and generate a draft release note.

• Agent Division of Labor: Follows the single responsibility principle, separating planning/execution/verification for easy maintenance and expansion.
• Natural Language Interface: A unified interaction entry that reduces the cognitive burden of multi-tool collaboration.
• Visual Feedback: Real-time process display builds human-machine trust and supports user intervention and adjustment.

Currently, AMOS mainly supports the GitHub-Jira-Slack combination. In the future, it will integrate tools like CI/CD, monitoring, and document collaboration; meanwhile, it will enhance the complex task planning capability of the Planner Agent.

AMOS represents the transformation of AI-assisted development from a single assistant to a multi-system orchestrator. It frees developers from tool switching, allowing teams to focus on creative problem-solving, and provides a reference paradigm for improving development efficiency.