JoySafeter: Enterprise AI Agent Platform for Security Automation

JoySafeter addresses the limitations of traditional security tooling by providing an AI-native platform for building, orchestrating, and running autonomous security agents at scale. It targets engineers and security professionals seeking to accelerate the deployment of sophisticated security automation, enabling complex tasks like vulnerability analysis and penetration testing to be executed in minutes rather than months. The platform facilitates a paradigm shift towards AI-driven Security Operations (AISecOps), leveraging multi-agent collaboration and dynamic adaptation to multiply capacity and replace brittle, manual processes.

How It Works

The platform employs an AI-native approach centered around multi-agent collaboration and visual orchestration. Core to its design are DeepAgents, which utilize a Manager-Worker architecture with evolving memory and a versioned skill system, all orchestrated via the LangGraph engine for stateful, graph-based workflows. This allows agents to dynamically adapt their actions based on real-time findings, a significant advantage over static scripts. Tool integration is streamlined through the Model Context Protocol (MCP), enabling seamless connection with over 200 tools without custom glue code.

Quick Start & Requirements

The recommended installation is via the ./deploy/quick-start.sh script, which offers an interactive menu for various Docker Compose and local development setups. Key prerequisites include Docker and Docker Compose. Detailed installation and production deployment guides are available in INSTALL.md, deploy/README.md, and deploy/PRODUCTION_IP_GUIDE.md.

Highlighted Details

• Visual Agent Builder: Features a no-code drag-and-drop workflow editor, a "Rapid Mode" for generating agents from natural language descriptions in minutes, and a "Deep Mode" for detailed visual debugging.
• Extensive Tooling: Integrates over 200 security tools (e.g., Nmap, Nuclei, Trivy) and supports custom extensions via the MCP Protocol, complemented by 30+ pre-built skills.
• DeepAgents Orchestration: Enables multi-level agent collaboration, long/short-term memory evolution for continuous learning, and a versioned skill system managed by the LangGraph engine.
• Enterprise Readiness: Offers multi-tenancy with role-based access control, comprehensive audit trails for compliance, SSO integration (GitHub, Google, Microsoft, OIDC), and isolated multi-tenant sandboxes for secure code execution.

Maintenance & Community

The project is developed by the "JoySafeter Team." Community support and discussion are available via a WeChat user group. Specific roadmap or contribution links beyond the repository itself are not detailed in the provided README.

Licensing & Compatibility

JoySafeter is licensed under the Apache License 2.0, which is generally permissive for commercial use and integration into closed-source projects. For specific commercial solutions or inquiries, contact JD Technology Solutions Team at org.ospo1@jd.com.

Limitations & Caveats

The provided README does not explicitly detail limitations, alpha status, or known bugs. Certain advanced functionalities, such as penetration testing scenarios, require a specific sandbox image (swr.cn-north-4.myhuaweicloud.com/ddn-k8s/ghcr.io/jd-opensource/joysafeter-sandbox:latest) to be configured.