What it does
The first open-source security scanner purpose-built for agentic AI systems. Analyses agent configurations, scans for insecure tool bindings, and simulates multi-agent attack scenarios. From Splx AI (now part of ZScaler).
Security relevance
Addresses a critical gap — most AI security tools focus on single-model interactions, but agentic systems introduce new attack surfaces: insecure tool use (LLM07), excessive agency (LLM08), and inter-agent manipulation. Agentic Radar scans for these agent-specific vulnerabilities. Maps directly to three OWASP Agentic Top 10 entries: ASI02 (Tool Misuse — scans for insecure tool bindings), ASI04 (Supply Chain — validates MCP server integrity), and ASI07 (Inter-Agent Communication — detects insecure agent-to-agent patterns). Now part of ZScaler's security portfolio.
When to use it
Use when deploying AI agents that use tools, access APIs, or interact with other agents. Essential for any MCP-based architecture or multi-agent system. Open-source and actively maintained, but requires Python setup and familiarity with agent architectures.
OWASP coverage
Risks addressed — mapped to both OWASP Top 10 standards. 2 in LLM, 3 in Agentic.
The raw record
What Yuntona stores. Single source of truth — fork it on GitHub.
name: Agentic Radar slug: agentic-radar type: Mixed category: AI Red Teaming url: https://splx.ai/resources/agentic-radar reviewed: 2026-04 added: 2026-04 updated: 2026-04 risks: llm: [LLM01, LLM07] asi: [ASI02, ASI04, ASI07] complexity: Guided Setup pricing: — audience: Red Team lifecycle: [develop] tags: [Agentic, Open Source, Scanner]