Compare orchestration topologies on one task family using integrity-gated observations and coordination metrics
Scanned 9/3/2026
Install to Claude Code
npx -y skills add jmagly/aiwg --skill topology-lab --agent claude-codeInstalls into .claude/skills of the current project.
Are you the author of Topology Lab?
Add the live security badge to your README — it updates automatically with every re-scan.
[](https://www.skillsdirectory.com/skills/jmagly-topology-lab)More formats (shields.io, HTML) on the badges page.
---
namespace: aiwg
name: topology-lab
platforms: [all]
description: Compare orchestration topologies on one task family using integrity-gated observations and coordination metrics
---
# Orchestration Topology Lab
Use this skill when deciding whether a task should remain single-agent or use a
bounded topology. Never select a topology from agent count alone.
1. Hold the task family, scoring contract, and integrity state constant.
2. Record observations for `single-agent`, `bounded-parallel`, and
`planner-worker`.
3. Run `aiwg topology-lab run <fixture.json>`.
4. Review quality, coordination overhead, contradiction rate, synthesis
failure, delegation precision, cost, and activity-evidence quality.
5. Treat missing activity evidence as `NOT RUN`, never as a zero or a pass.
The result is an experimental recommendation. Production routing remains out of
scope until repeated trusted evaluations justify policy.
@implements #2042
Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.
No comments yet. Be the first to comment!
Adversarial multi-agent planning skill. Self-orchestrates 5 hostile category members (unspecified-low, unspecified-high, deep, ultrabrain, artistry) via team-mode for ruthless cross-critique debate, distills only the defensible insights, then MANDATORILY hands the distilled insight bundle to the `plan` agent for executable plan formalization. Use when planning needs maximum rigor and surfacing of weak assumptions, blind spots, and over-engineering. Triggers: 'hyperplan', 'hpp', '/hyperplan', ...