This project offers an enterprise-grade, open-source IT operations (ITOps) automation platform powered by large language models (LLMs) and a multi-agent architecture. It targets IT professionals seeking to automate server management, incident response, and diagnostic tasks through a visual workflow engine, aiming to close the loop on Zabbix/Prometheus alerts with AI-driven solutions.

How It Works

The platform orchestrates multiple AI agents via a visual workflow editor, enabling users to build automated pipelines. It integrates with various LLMs (cloud and local) to power agent capabilities like root cause analysis, automated remediation command generation, and natural language querying. Data flows from monitoring systems (Zabbix, Prometheus) via webhooks into the platform, where agents process alerts, execute commands via SSH, and provide notifications. Core components include a React frontend, Node.js/Express backend, SQLite database, and WebSocket for real-time communication.

Quick Start & Requirements

• Primary install: One-click deployment scripts (deploy.ps1 for Windows, deploy.sh for Linux/Mac) or Docker Compose.
• Prerequisites: Docker, Docker Compose. No specific GPU or CUDA requirements are listed for the platform itself, but local LLM deployment would necessitate them.
• Estimated setup: Approximately 5 minutes for Docker deployment.
• Links:
  • Demo Environment: https://www.zjzwfw.cloud/ITOpsAgentinfo (Project Website)
  • Documentation: docs/ (within the repository)

Highlighted Details

• Multi-Agent Collaboration: Features 9 pre-set operational agents (alerting, diagnostics, inspection, etc.) with support for custom agent creation.
• Visual Workflow Orchestration: Drag-and-drop interface for creating serial, parallel, or conditional workflows with real-time progress updates.
• AI-Driven Automation: Enables AI-powered root cause analysis, automated remediation command generation, and AI Copilot for natural language interaction.
• Broad LLM Support: Integrates with a model pool including Doubao, DeepSeek, Qwen, OpenAI, Zhipu, and local models via Ollama, LM Studio, or vLLM.
• Enterprise Security: Implements AES-256-GCM encryption for credentials, JWT authentication, rate limiting, audit logs, and non-root container execution.

Maintenance & Community

The project is primarily maintained by a single author, Tan Ce (@qinshihu). Community contribution is encouraged via bug reports, feature requests, and documentation improvements. Links to a blog and WeChat public account are provided for the author.

Licensing & Compatibility

• License: Mozilla Public License 2.0 (MPL-2.0) for code committed after May 27, 2026. Code committed before this date follows the MIT license.
• Restrictions: The project explicitly prohibits closed-source development, repackaging for sale, or SaaS operations. It is committed to remaining permanently open-source.

Limitations & Caveats

The project's license (MPL-2.0) requires derivative works to also be open-sourced under MPL-2.0, which may impact commercial adoption strategies. While it supports numerous LLMs, optimal performance for AI features may depend on the chosen model and underlying hardware.