OpenDeRisk is an AI-native system designed for application risk intelligence and root cause analysis (RCA). It targets Site Reliability Engineers (SREs) and developers, offering automated, in-depth diagnostics and visualized evidence chains for complex operational issues.

How It Works

OpenDeRisk employs a multi-agent architecture, orchestrating specialized agents (SRE-Agent, Code-Agent, ReportAgent, Vis-Agent, Data-Agent) to collaboratively analyze logs, traces, and code. A key innovation is the Code-Agent's ability to dynamically write code for real-time analysis, enhancing diagnostic depth. The system visualizes the entire diagnostic process and evidence chain using the Vis protocol, promoting transparency and rapid judgment.

Quick Start & Requirements

• Install: curl -LsSf https://astral.sh/uv/install.sh | sh followed by uv sync --all-packages --frozen --extra "base" --extra "proxy_openai" --extra "rag" --extra "storage_chromadb" --extra "client".
• Prerequisites: API keys (e.g., DeepSeek), Python, and the ~26GB OpenRCA dataset (Telecom subset provided via gdown).
• Setup: Requires downloading and configuring the dataset.
• Docs: English, 中文, Video Tutorial.

Highlighted Details

• Utilizes a multi-agent system for collaborative RCA.
• Features a Code-Agent that dynamically writes code for analysis.
• Provides visualized evidence chains for diagnostic transparency.
• Built on a completely open-source architecture.

Maintenance & Community

Acknowledges contributions from DB-GPT, GPT-Vis, MetaGPT, and OpenRCA. Community engagement is encouraged via a Dingding group.

Licensing & Compatibility

The README does not explicitly state the license. Compatibility for commercial use or closed-source linking is not specified.

Limitations & Caveats

The project relies on a large dataset (~26GB) and requires specific API keys. The README does not specify the project's license, which could impact commercial adoption.