/security/

Perspective

Security — the attack surface of a multi-agent system.

How multi-agent systems get attacked, and which defenses close each gap. Two OWASP catalogs — the LLM Top 10 and the Agentic Threats — mapped to the guardrails, least-privilege, and human gates this project already documents.

5
Attack surfaces
27
Threats
2
OWASP catalogs
10
ASI patterns

Six ways a MAS amplifies risk a single call never carries.

Adding agents does not add attack surface linearly. Several risks below exist only once agents coordinate, and each raises the cost of one compromise well past anything a lone LLM call could expose.

  • 01

    Blast radius

    A single LLM call fails in isolation. A compromised agent can pass its corrupted state to every agent that trusts its output without re-verifying it, so one exploited hand-off cascades into a system-wide failure.

  • 02

    Agent collusion

    Two or more compromised or subtly misaligned agents can coordinate — through shared memory, negotiated protocol messages, or repeated interaction — to manipulate a decision in a way no individual agent's output would flag as anomalous.

  • 03

    Identity sprawl

    One agent needs one identity. A dozen agents, each holding its own tool credentials and delegated permissions, turn identity and access management into a combinatorial problem — every added identity is one more credential to spoof or over-provision.

  • 04

    Coordination failures in dynamic environments

    The routing, voting, or consensus logic that lets agents adapt to changing conditions is rarely tested against adversarial conditions, and can break down — or be driven into breaking down — once an adversary controls part of that variance.

  • 05

    Decision-lineage & auditability gaps

    No single agent holds the full reasoning trace behind a multi-agent decision, so reconstructing why the system acted means stitching together partial, differently-shaped traces from every agent that touched it.

  • 06

    Man-in-the-middle on inter-agent channels

    The messages agents exchange — hand-offs, blackboard writes, negotiated votes — travel over a channel a single-LLM application never has, and an attacker who intercepts or alters it changes what one agent believes another said.

THE MAP

All 27 threats, located on the architecture.

Before the index lists them, the threat map shows where each threat enters — from untrusted input through the agent runtime to the triggered action. Hover a threat to see where it lands and which further surfaces it touches.

☁ Untrusted mediumUsers & attackersPrompts + context from an untrusted source. CRUD operations.
☁ Untrusted mediumExternal data sourcesWeb, documents, third-party APIs, shared repos & model registries — grounding, retrieval & supply chain from the open internet.
TB · Ingress
01Input6
User prompts & ingested untrusted content
API gateway
Auth · Rate-limit · Routing
Input guardrail
Prompt-guard · Schema check
Threats at this surface
TB01 · Agent runtime
02Agent runtime6
Supervisor delegates to specialists & merges results
Supervisor / Orchestrator
plans · delegates · merges
A2A · Message busAgent #2
Worker · Execution loop · Input → Output (NL / Media)
PlanningActionTool / Function-calling
LLM model · Function-calling
augmented model
Blackboard / Shared state
shared working state
Threats on the inter-agent surface
TB02 · Knowledge store
03Memory3
Short- & long-term state the agents rely on
Working memory
Scratchpad · Task state
Vector store (RAG)
Long-term · Retrieval
Threats at this surface
TB03 · Tool layer
04Tools & services6
The capabilities the agents call & their supply chain
Tool registry
discover · bind
MCP server
external tool providers
Sandbox execution
isolated code execution
Threats at this surface
TB · Action & oversight
05Output6
Generated output, triggered action & the oversight over it
Output guardrail
LLM-as-Judge · Validation
HITL gate
human approval
Audit trail
Logging · Distributed tracing
Threats at this surface
OWASP LLM Top 10OWASP Agentic Threats (T1–T17)Trust boundaryUntrusted medium●● = touches further attack surfaces
The index

27 threats, ordered by attack surface.

Rather than two separate catalogs, every entry of the OWASP LLM Top 10 and the Agentic Threats (T1–T17) is filed under the surface it enters. Threats spanning several surfaces appear more than once — each row keeps its origin tag (LLM / Agentic) and its other surfaces.

01

Input / Prompt

6 threats

Everything the system reads — user prompts and untrusted content an agent ingests — where injection enters.

02

Memory / State

5 threats

Persistent state and retrieved context, where poisoned memory biases every later decision.

03

Tools & External Data

11 threats

The tools and external data an agent calls, where misuse and poisoned results turn reach into risk.

04

Inter-Agent Communication

9 threats

Messages between agents, where a poisoned or spoofed peer propagates compromise across the system.

05

Output / Actuation

11 threats

What the system emits or actuates, where unvalidated output and excessive agency cause real-world harm.

Agentic AI Patterns

The OWASP Agentic Security Initiative's shared vocabulary for threat-modeling conversations. Each maps to a pattern this project already covers in depth.

  • Reflective Agent

    Agents that iteratively evaluate and critique their own outputs to enhance performance.

    AI code generators that review and debug their own outputs, like Codex with self-evaluation.

    Maps to: Reflexion
  • Task-Oriented Agent

    Agents designed to handle specific tasks with clear objectives.

    Automated customer-service agents for appointment scheduling or returns processing.

    Maps to: ReAct
  • Hierarchical Agent

    Agents organized in a hierarchy, managing multi-step workflows or distributed control systems.

    Project-management systems where higher-level agents oversee task delegation.

    Maps to: Hierarchical Supervisor
  • Coordinating Agent

    Agents facilitate collaboration, coordination, and tracking, ensuring efficient execution.

    A coordinator assigns subtasks to specialists in an AI-powered DevOps workflow: one plans deployments, another monitors performance, a third handles rollbacks.

    Maps to: Orchestrator-Workers
  • Distributed Agent Ecosystem

    Agents interact within a decentralized ecosystem, often in IoT or marketplaces.

    Autonomous IoT agents managing smart-home devices, or a marketplace with buyer and seller agents.

    Maps to: Swarm / Contract-Net
  • Human-in-the-Loop Collaboration

    Agents operate semi-autonomously with human oversight.

    AI-assisted medical-diagnosis tools that recommend but let doctors make the final decision.

    Maps to: HITL Gate
  • Self-Learning and Adaptive Agents

    Agents adapt through continuous learning from interactions and feedback.

    Co-pilots that adapt to user interactions over time, learning from feedback.

    Maps to: Skill-Build / Reflector
  • RAG-Based Agent

    Agents use Retrieval-Augmented Generation to draw on external knowledge sources dynamically.

    Agents performing real-time web browsing for research assistance.

    Maps to: Agentic RAG
  • Planning Agent

    Agents autonomously devise and execute multi-step plans to achieve complex objectives.

    Task-management systems organizing and prioritizing tasks by user goals.

    Maps to: Plan-and-Execute
  • Context-Aware Agent

    Agents dynamically adjust their behavior and decision-making based on the context in which they operate.

    Smart-home systems adjusting settings based on user preferences.

    Maps to: Virtual Context Management

Search

Search patterns, frameworks, and pages.