Hallucinated Routing

Observable symptom: Reflection loops never converge. Recursion limits trip at runtime instead of design time. Suitable correction: Schema-validate router decisions. Hard recursion_limit , hard max_handoffs .

Use this profile to understand the building block briefly, place it in the model, and switch to the 360° assessment when needed.

Content type

Concept

Theoretical construct: explains a term, principle, or mental model.

Classification level

Structure

What organizes, connects, or makes decisions possible.

Available perspectives

Open this building block directly in an already curated view.

360°

In brief

Hallucinated routing in brief

Hallucinated routing occurs when a multi-agent system assumes a nonexistent or unsuitable agent path and builds on it.

Where does it come from?

The pattern is described as an anti-pattern in the analysis of multi-agent systems. It arises from unverified assumptions about roles, capabilities, or reachable agents rather than a checked topology.

The mental model

An agent chooses a recipient for a task. Without checking the actual route, it may select a plausible but invalid recipient. The chain then produces errors, loops, or seemingly reasoned results. An explicit agent registry, capability metadata, and handoff protocol make the path testable.

Key building blocks

Implicit assumption

An agent treats a presumed role or capability as available without checking it at runtime.

Agent topology

Topology records which agents exist, are reachable, and may receive a handoff.

Route validation

Registries, capability checks, and protocols prevent invalid or unbounded handoffs.

When is it useful?

The pattern matters for reliability, debugging, and security in agentic systems. Make handoffs observable and handle failures explicitly instead of treating plausible agent output as proof of correct routing.

Multi Agent System Education – Hallucinated Routing