AI Engineering5 min readchapterQuick Read
Multi-Agent Topologies and Collaboration Contracts
Select and govern multiple agents only when specialization, parallelism, or independent assurance creates measurable value.
Status: Review readyRisk: highLifecycle: plan · execute · verify · learnContent reviewed 2026-08-30Maturity guide →
Claim boundaryThis is curriculum guidance. It does not by itself prove a production implementation.
study mode
A rapid review of the chapter’s existing Quick Read, principles, definitions, lessons, and review material.
Quick Read
~2 minThe chapter in one pass.
- Purpose: Select and govern multiple agents only when specialization, parallelism, or independent assurance creates measurable value.
- Default: A deterministic workflow or one bounded agent is simpler and often safer.
- Critical contracts: Delegation, authority, context, handoff, completion, disagreement, independence, budget, failure, and evidence lineage.
- Anti-pattern: Several agents using the same model, prompt, context, and assumptions do not constitute independent verification.
11. Review exercise
Design a security-sensitive change using both a single-agent workflow and a supervisor with builder and independent critic. Define the expected measurable benefit, inject one correlated failure and one partial child result, compare quality, latency, cost, and review burden, then choose the minimum sufficient design.
External review
Review this chapter.
Challenge a claim, boundary, missing failure mode, unclear term, or unsupported evidence statement.
- Claim
- Boundary
- Failure
- Evidence