Team Topology Analysis

Description

Provide framework for analyzing team structure rationality, used to evaluate whether collaboration topology between agents is efficient and whether bottlenecks or redundancies exist.

Belongs To

This skill belongs exclusively to agents/discovery/role-designer.md

Core Knowledge

Topology Patterns

1. Pipeline Topology

code

A → B → C → D

•Applicable scenario: Work has strict sequential order
•Advantages: Clear responsibilities, predictable process
•Risk: Any blocked node causes entire chain to stall

2. Hub & Spoke Topology

code

    B
    ↑
A ← Coordinator → C
    ↓
    D

•Applicable scenario: Multiple parallel work streams coordinated by one coordinator
•Advantages: Flexible assignment, easy to add/remove nodes
•Risk: Coordinator becomes single point of bottleneck

3. Hybrid Topology

code

Coordinator → [GroupA: A1 → A2] → Review → [GroupB: B1, B2] → Final

•Applicable scenario: Complex teams with some sequential work, some parallel
•Usually the recommended final pattern

Topology Health Check

•Bottleneck detection: Is any agent the upstream of 3+ other agents simultaneously?
•Island detection: Is any agent without any interaction with other team members?
•Cycle detection: Does A→B→C→A circular dependency exist? (review cycles excluded)
•Coordinator load: Does coordinator directly manage more than 7 agents? If so, recommend splitting into separate teams.

Agent Teams Mode Considerations

Shared Task List Patterns

•Use task dependencies (blockedBy) to enforce sequential gates
•Parallel agents claim tasks independently — no coordinator bottleneck
•Task ownership prevents duplicate work on the same deliverable

File Ownership Rules

•Parallel agents must write to non-overlapping file sets
•Define ownership in each agent's "Communication Patterns" section
•On conflict: coordinator mediates, or designate a primary owner

Communication Decision Framework

Scenario	Channel	Rationale
Agent needs input from specific peer	Direct message	Avoids noise for others
Architecture decision change	Broadcast	Affects all agents
Task reassignment or priority change	Via coordinator	Coordinator tracks assignments
Review feedback	Direct message	Only reviewer ↔ author

Coordinator Bottleneck Mitigation

•Allow peer-to-peer communication for review pairs
•Limit broadcast to 2-3 defined critical triggers
•Coordinator focuses on: task assignment, dependency resolution, final quality gate
•Avoid routing routine status updates through coordinator

Quality Checkpoints

• Every agent has at least one upstream or downstream connection
• Coordinator is at the logical center of topology
• No meaningless circular dependencies
• Parallel work agents are correctly grouped
• (Agent Teams) File ownership is non-overlapping between parallel agents
• (Agent Teams) Communication channels are defined for every peer-to-peer pair
• (Agent Teams) Broadcast triggers are limited and clearly defined