Version updated for https://github.com/taoq-ai/ziran to version v0.30.0. This action is used across all versions by 1 repositories. Action Type This is a Composite action. Go to the GitHub Marketplace to find the latest changes. Action Summary ZIRAN is a GitHub Action designed to identify and test vulnerabilities in AI agents, including those with tools, memory, and multi-step reasoning. It automates security testing by modeling agents as interconnected graphs of capabilities, enabling the detection of vulnerabilities that arise from the interaction of tools and features, which traditional isolated testing methods often miss. Key capabilities include graph-based tool chain discovery, side-effect detection, autonomous pentesting, and multi-agent coordination, providing a comprehensive approach to AI agent security.

Version updated for https://github.com/taoq-ai/ziran to version v0.28.0. This action is used across all versions by 1 repositories. Action Type This is a Composite action. Go to the GitHub Marketplace to find the latest changes. Action Summary ZIRAN is a security testing tool designed to identify vulnerabilities in AI agents, particularly those using tools, memory, and multi-step reasoning. It models agents as capability graphs to uncover risks arising from tool interactions, such as transitive attack paths and silent failures, which are often missed by traditional prompt-based testing. Key features include graph-based tool chain analysis, execution-level side-effect detection, multi-phase attack simulations, and support for autonomous pentesting and multi-agent coordination.

Version updated for https://github.com/taoq-ai/ziran to version v0.26.0. This action is used across all versions by 1 repositories. Action Type This is a Composite action. Go to the GitHub Marketplace to find the latest changes. Action Summary ZIRAN is a GitHub Action designed for security testing of AI agents, including those with tools, memory, and multi-step reasoning capabilities. It automates the detection of vulnerabilities by modeling agents as capability graphs to identify risks arising from tool interactions and multi-phase attack campaigns, which traditional prompt-based testing often overlooks. Key features include tool chain analysis, execution-level side-effect detection, autonomous pentesting, and multi-agent coordination, making it a comprehensive solution for assessing and mitigating security risks in AI systems.

Version updated for https://github.com/taoq-ai/ziran to version v0.25.0. This action is used across all versions by 1 repositories. Action Type This is a Composite action. Go to the GitHub Marketplace to find the latest changes. Action Summary ZIRAN is a GitHub Action designed for security testing of AI agents, focusing on identifying vulnerabilities arising from the interaction of tools, memory, and multi-step reasoning within agents. It automates the discovery of complex attack vectors, such as transitive attack paths and side-effect vulnerabilities, using a graph-based analysis approach to model agent capabilities. ZIRAN also enables execution-level monitoring, autonomous penetration testing, and multi-agent coordination, providing comprehensive security assessments that go beyond traditional prompt or tool isolation testing.