Back to Risk Catalogue

Agent Threat Rules (ATR)

Open detection rule format for AI agent threats, MIT licensed. Rules map to MITRE ATLAS, OWASP LLM Top 10, OWASP Agentic AI Top 10, and NIST AI RMF.

ATR-2026-00010 Malicious Content in MCP Tool Response

Risks

RI-10Prompt Injection RI-25Tool Chain Manipulation and Injection RI-26MCP Server Supply Chain Compromise

Mitigations

MI-20MCP Server Security Governance

ATR-2026-00011 Instruction Injection via Tool Output

Risks

RI-10Prompt Injection RI-25Tool Chain Manipulation and Injection

Mitigations

MI-19Tool Chain Validation and Sanitization

ATR-2026-00012 Unauthorized Tool Call

Risks

RI-24Agent Action Authorization Bypass RI-25Tool Chain Manipulation and Injection

Mitigations

MI-18Agent Authority Least Privilege Framework MI-19Tool Chain Validation and Sanitization

ATR-2026-00013 Tool-Driven Server-Side Request Forgery

Risks

RI-25Tool Chain Manipulation and Injection

Mitigations

MI-19Tool Chain Validation and Sanitization

ATR-2026-00030 Cross-Agent Attack Detection

Risks

RI-28Multi-Agent Trust Boundary Violations

Mitigations

MI-22Multi-Agent Isolation and Segmentation

ATR-2026-00040 Agent Privilege Escalation

Risks

RI-24Agent Action Authorization Bypass

Mitigations

MI-18Agent Authority Least Privilege Framework

ATR-2026-00075 Agent Memory Manipulation

Risks

RI-27Agent State Persistence Poisoning

ATR-2026-00076 Insecure Inter-Agent Communication

Risks

RI-28Multi-Agent Trust Boundary Violations

Mitigations

MI-22Multi-Agent Isolation and Segmentation

ATR-2026-00098 Unauthorized Financial Action by Agent

Risks

RI-24Agent Action Authorization Bypass

Mitigations

MI-18Agent Authority Least Privilege Framework

ATR-2026-00113 Credential Theft via Agent Channel

Risks

RI-29Agent-Mediated Credential Discovery and Harvesting

Mitigations

MI-23Agentic System Credential Protection Framework

ATR-2026-00118 Human Approval Fatigue Exploitation

Risks

RI-24Agent Action Authorization Bypass