ATR Detection Rule Mapping for Agentic AI Top 10

June 2, 2026 ยท View on GitHub

Agent Threat Rules (ATR) is an open-source set of detection rules for AI agent security threats. This document maps ATR's 71 rules to each Agentic AI Top 10 vulnerability category, providing practitioners with executable detection for each threat.

ATR version: v0.4.0 (71 rules) | License: MIT | Engines: TypeScript + Python

Coverage Summary

AAIVulnerabilityATR RulesCoverage
AAI001Agent Authorization and Control Hijacking22 rulesSTRONG
AAI002Agent Tool Interaction Manipulation11 rulesSTRONG
AAI003Agent Goal and Instruction Manipulation13 rulesSTRONG
AAI004Agent Hallucination Exploitation4 rulesMODERATE
AAI005Agent Impact Chain and Blast Radius11 rulesSTRONG
AAI006Agent Memory and Context Manipulation8 rulesSTRONG
AAI007Agent Orchestration and Multi-Agent Exploitation12 rulesSTRONG
AAI008Agent Resource and Service Exhaustion4 rulesMODERATE
AAI009Agent Supply Chain and Dependency Attacks8 rulesSTRONG
AAI010Agent Knowledge Base Poisoning8 rulesSTRONG
AAI011Agent Untraceability8 rulesSTRONG
AAI012Checker-out-of-the-Loop Vulnerability5 rulesMODERATE

Detailed Mappings

AAI001: Agent Authorization and Control Hijacking

Prompt injection and privilege escalation attacks that hijack agent behavior.

ATR RuleTitleSeverity
ATR-2026-001Direct Prompt InjectionCRITICAL
ATR-2026-002Indirect Prompt InjectionCRITICAL
ATR-2026-003Jailbreak AttemptHIGH
ATR-2026-004System Prompt OverrideCRITICAL
ATR-2026-005Multi-Turn InjectionHIGH
ATR-2026-040Privilege EscalationCRITICAL
ATR-2026-041Scope CreepHIGH
ATR-2026-080Encoding EvasionMEDIUM
ATR-2026-081Semantic Multi-TurnHIGH
ATR-2026-091Nested PayloadHIGH
ATR-2026-093Gradual EscalationHIGH
ATR-2026-097CJK Injection PatternsMEDIUM
ATR-2026-104Persona HijackingHIGH
ATR-2026-107Delayed Execution BypassHIGH
ATR-2026-113Credential TheftCRITICAL
ATR-2026-114OAuth Token AbuseHIGH
ATR-2026-115Env Var HarvestingCRITICAL

AAI002: Agent Tool Interaction Manipulation

Agents bending tools into destructive outputs or using tools beyond intended scope.

ATR RuleTitleSeverity
ATR-2026-010MCP Malicious ResponseCRITICAL
ATR-2026-011Tool Output InjectionHIGH
ATR-2026-012Unauthorized Tool CallHIGH
ATR-2026-013Tool SSRFCRITICAL
ATR-2026-100Consent Bypass InstructionHIGH
ATR-2026-101Trust Escalation OverrideHIGH
ATR-2026-103Hidden Safety Bypass InstructionCRITICAL
ATR-2026-105Silent Action ConcealmentHIGH
ATR-2026-106Schema-Description ContradictionMEDIUM

AAI003: Agent Goal and Instruction Manipulation

Attacks that manipulate agent goals and instructions through injected directives.

ATR RuleTitleSeverity
ATR-2026-001Direct Prompt InjectionCRITICAL
ATR-2026-002Indirect Prompt InjectionCRITICAL
ATR-2026-003Jailbreak AttemptHIGH
ATR-2026-004System Prompt OverrideCRITICAL
ATR-2026-005Multi-Turn InjectionHIGH
ATR-2026-032Goal HijackingCRITICAL
ATR-2026-084Structured Data InjectionMEDIUM
ATR-2026-086Visual SpoofingMEDIUM
ATR-2026-091Nested PayloadHIGH
ATR-2026-093Gradual EscalationHIGH
ATR-2026-097CJK Injection PatternsMEDIUM
ATR-2026-104Persona HijackingHIGH

AAI004: Agent Hallucination Exploitation

Exploiting hallucinated outputs for cascading impact.

ATR RuleTitleSeverity
ATR-2026-050Runaway Agent LoopHIGH
ATR-2026-052Cascading FailureCRITICAL
ATR-2026-072Model Behavior ExtractionHIGH
ATR-2026-073Malicious Finetuning DataHIGH

AAI005: Agent Impact Chain and Blast Radius

Cascading effects across interconnected agent systems.

ATR RuleTitleSeverity
ATR-2026-030Cross-Agent AttackHIGH
ATR-2026-050Runaway Agent LoopHIGH
ATR-2026-051Resource ExhaustionHIGH
ATR-2026-052Cascading FailureCRITICAL
ATR-2026-063Skill Chain AttackHIGH
ATR-2026-074Cross-Agent Privilege EscalationCRITICAL
ATR-2026-076Inter-Agent Message SpoofingHIGH
ATR-2026-108Consensus Sybil AttackHIGH
ATR-2026-117Agent Identity SpoofingCRITICAL

AAI006: Agent Memory and Context Manipulation

Memory poisoning that reshapes agent behavior long after the initial interaction.

ATR RuleTitleSeverity
ATR-2026-020System Prompt LeakHIGH
ATR-2026-070Data PoisoningHIGH
ATR-2026-075Agent Memory ManipulationHIGH
ATR-2026-085Audit EvasionHIGH
ATR-2026-090Threat Intel ExfilHIGH
ATR-2026-092Consensus PoisoningHIGH
ATR-2026-094Audit BypassHIGH
ATR-2026-073Malicious Finetuning DataHIGH

AAI007: Agent Orchestration and Multi-Agent Exploitation

Attacks targeting multi-agent systems and inter-agent communication.

ATR RuleTitleSeverity
ATR-2026-030Cross-Agent AttackHIGH
ATR-2026-032Goal HijackingCRITICAL
ATR-2026-041Scope CreepHIGH
ATR-2026-074Cross-Agent Privilege EscalationCRITICAL
ATR-2026-076Inter-Agent Message SpoofingHIGH
ATR-2026-089Polymorphic SkillHIGH
ATR-2026-107Delayed Execution BypassHIGH
ATR-2026-108Consensus Sybil AttackHIGH
ATR-2026-116A2A Message ValidationHIGH
ATR-2026-117Agent Identity SpoofingCRITICAL

AAI008: Agent Resource and Service Exhaustion

Resource exhaustion and denial-of-service through agent automation.

ATR RuleTitleSeverity
ATR-2026-050Runaway Agent LoopHIGH
ATR-2026-051Resource ExhaustionHIGH
ATR-2026-052Cascading FailureCRITICAL
ATR-2026-088Adaptive CountermeasureHIGH

AAI009: Agent Supply Chain and Dependency Attacks

Poisoning of dynamic MCP and A2A ecosystems at runtime.

ATR RuleTitleSeverity
ATR-2026-060Skill ImpersonationCRITICAL
ATR-2026-061Description-Behavior MismatchHIGH
ATR-2026-062Hidden CapabilityHIGH
ATR-2026-063Skill Chain AttackHIGH
ATR-2026-065Skill Update AttackHIGH
ATR-2026-066Parameter InjectionHIGH
ATR-2026-089Polymorphic SkillHIGH
ATR-2026-095Supply Chain PoisoningCRITICAL

AAI010: Agent Knowledge Base Poisoning

Contaminating the knowledge sources that agents rely on.

ATR RuleTitleSeverity
ATR-2026-070Data PoisoningHIGH
ATR-2026-073Malicious Finetuning DataHIGH
ATR-2026-075Agent Memory ManipulationHIGH
ATR-2026-092Consensus PoisoningHIGH
ATR-2026-096Registry PoisoningHIGH

AAI011: Agent Untraceability

Evasion of audit trails and accountability mechanisms.

ATR RuleTitleSeverity
ATR-2026-082Fingerprint EvasionMEDIUM
ATR-2026-085Audit EvasionHIGH
ATR-2026-087Rule ProbingMEDIUM
ATR-2026-088Adaptive CountermeasureHIGH
ATR-2026-094Audit BypassHIGH
ATR-2026-102Disguised Analytics ExfiltrationHIGH
ATR-2026-105Silent Action ConcealmentHIGH

AAI012: Checker-out-of-the-Loop Vulnerability

Exploiting lack of human oversight in automated agent decisions.

ATR RuleTitleSeverity
ATR-2026-077Human Trust ExploitationHIGH
ATR-2026-098Unauthorized Financial ActionCRITICAL
ATR-2026-099High-Risk Tool GateMEDIUM
ATR-2026-118Approval Fatigue ExploitationMEDIUM
ATR-2026-119Social Engineering via AgentHIGH

How to Use ATR Rules

Quick scan (CLI)

npx agent-threat-rules scan your-mcp-config.json

TypeScript

import { ATREngine } from 'agent-threat-rules';
const engine = new ATREngine();
const results = engine.scan(content);

Python

from atr import ATREngine
engine = ATREngine()
results = engine.scan(content)

SIEM Integration

atr convert splunk --output atr-rules.spl
atr convert elastic --output atr-rules.json

References