Name: swarm-safety
Availability: InStock
Author: swarm-ai-safety

System Documentation

What problem does it solve?

This Skill helps you understand and mitigate emergent risks in multi-agent AI systems by simulating complex interactions and evaluating safety through novel metrics.

Core Features & Use Cases

Simulate AI Swarms: Run complex multi-agent simulations with diverse agent types (honest, deceptive, adversarial) and configurable governance mechanisms (taxes, audits, circuit breakers).
Measure Distributional Safety: Utilize soft probabilistic labels and continuous metrics like toxicity, quality gap, and incoherence to detect subtle failure modes.
Use Case: Test how different governance strategies perform against adaptive adversaries in a simulated marketplace, identifying critical thresholds before real-world deployment.

Quick Start

Run the 'ai_economist_full' scenario with default settings.

Dependency Matrix

Required Modules

numpypydanticpandaspyyamlrequeststenacitymatplotlibseabornscipypytestruffmemory_profilerpsutilmkdocs-materialpymdown-extensionsmkdocs-git-revision-date-localized-pluginmkdocs-rss-pluginfastapiuvicorn[standard]python-multiparthttpxanthropicopenaigoogle-genaillama-cpp-pythonstreamlitplotlycrewailanggraphlangchain-corelangchain-anthropiclangchain-ollamalangchain-openailangchain-text-splitterschromadbleanngamescapedarwinian-evolver @ git+https://github.com/imbue-ai/darwinian_evolver.gitprimeverifiersletta-clienthodoscopedockerrich

Components

scriptsreferences

Please help me install this Skill: Name: swarm-safety Download link: https://github.com/swarm-ai-safety/swarm/archive/main.zip#swarm-safety Please download this .zip file, extract it, and install it in the .claude/skills/ directory.

swarm-safety