skill-agent-topology

1> **Host: Codex CLI** — This skill was designed for Claude Code and adapted for Codex.
2> Cross-reference commands use installed skill names in Codex rather than `/octo:*` slash commands.
3> Use the active Codex shell and subagent tools. Do not claim a provider, model, or host subagent is available until the current session exposes it.
4> For host tool equivalents, see `skills/blocks/codex-host-adapter.md`.
5 
6 
7# Agent Topology Audit
8 
9Most advice about multi-agent systems is about how to add agents. This is about
10whether to. It audits a setup you already have, counts what each boundary
11between agents costs, and compares that against what the boundary buys. Removing
12an agent is a valid, and often the correct, result.
13 
14The framing comes from Liu, Canhui (2026), *The Organizational Behavior of
15Agentic AI* ([arXiv:2606.30986](https://arxiv.org/abs/2606.30986)), which models
16coordination overhead as **contextual transaction cost** — the cost of making
17task context usable across an agent boundary.
18 
19## When To Use
20 
21- Before adding another agent, seat, or phase to a workflow that already works.
22- When a workflow is slow and it is not obvious which part is earning its time.
23- When agents keep agreeing. Agreement that costs three dispatches and produces
24  what one would have produced is overhead wearing the costume of consensus.
25- When a handoff keeps losing something and the fix keeps being "add more
26  context to the prompt".
27…