st-lzh/Wuhr-AI-ops

What it solves

Wuhr AI Ops provides a unified, auditable control plane for IT operations teams to manage servers, network devices, Kubernetes clusters, and CI/CD pipelines. It replaces simple AI chat interfaces with a closed-loop system that combines natural language analysis with real-world command execution, human approval workflows, and persistent execution records to ensure safety and traceability in infrastructure management.

How it works

The system consists of a Docker-based management platform and a lightweight Agent installed directly on target operating systems. Users interact with an AI assistant to specify targets (using @ mentions for hosts, IPs, or groups) and goals. The AI generates an execution plan; for high-risk actions, the platform requires human approval before the Agent executes the commands. The platform supports multi-model integration and MCP (Model Context Protocol) tools, ensuring that API keys and credentials are encrypted and stored securely on the server side rather than in the browser.

Who it’s for

It is designed for trusted operations teams who need to automate routine maintenance, network configuration, and application delivery while maintaining strict security boundaries and audit logs.

Highlights

• Closed-Loop Execution: Integrates AI planning, command execution, human approval, and result summarization into one traceable workflow.
• Unified Resource Management: Single entry point for managing Linux hosts, Kubernetes, switches, routers, and firewalls.
• Multi-Target Efficiency: Generates a single plan that can be reused across multiple target machines to reduce token consumption.
• Integrated Delivery: Combines CI/CD pipelines (Jenkins integration) with AI-driven diagnostics for build failures and deployment anomalies.
• Security-First Design: Features privileged command blocking, encrypted credential storage, and mandatory human-in-the-loop approvals for risky changes.