ziran

What problem does it solve?

This Skill enables automated identification of security risks within AI agents by analyzing their capabilities, tools, and interactions.

Core Features & Use Cases

• Tool Chain Mapping: Automatically constructs a capability graph to detect dangerous tool sequences.
• Multi-Phase Campaign Simulation: Runs multi-step exploits to evaluate agent safety under varied scenarios.
• Security Event Detection: Identifies side-effects, prompt injections, and unsafe behavior at both the text and execution levels.
• Use Case: Security teams can simulate attack campaigns on proprietary AI agents to find vulnerabilities before deployment.

Quick Start

Run the ZIRAN security analysis against your AI agent by providing its source code or API endpoint to identify potential security issues.