sanity-check

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

Validate multi-agent simulation scenario configs with short runs and metric checks.

39|4|Updated Feb 3, 2026

Search Tags:#validation #multi-agent #simulation #risk assessment #sanity test #scenario check

Ready to run

npx skills add https://github.com/swarm-ai-safety/swarm --skill sanity-check-swarm-ai-safety

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: sanity-check
Source: https://github.com/swarm-ai-safety/swarm/tree/main/.skills/sanity-check
Command: npx skills add https://github.com/swarm-ai-safety/swarm --skill sanity-check-swarm-ai-safety

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) and references (resource) components.

What problem does it solve?

This Skill quickly validates a scenario configuration before committing to a lengthy simulation run, preventing wasted computational resources and identifying potential issues early.

Core Features & Use Cases

Scenario Validation: Checks for the existence and basic YAML structure of scenario files.
Short Run Execution: Performs a minimal simulation run to test for crashes and basic metric calculation.
Metric Sanity Check: Verifies that key metrics fall within expected ranges (e.g., probabilities between 0 and 1, finite values).
Use Case: Before launching a multi-hour simulation sweep across hundreds of seeds, use this Skill to run a single seed for a few epochs to ensure the scenario file is correctly formatted and the simulation doesn't immediately error out.

Quick Start

Run a sanity check on the 'basic_scenario' scenario.

Frequently Asked Questions about sanity-check

High-intent search queries and answers about installing and using this skill.

FAQPage Schema

How do I validate a multi-agent simulation scenario before running a full sweep?▼

To validate a multi-agent simulation scenario, run a quick sanity test that checks YAML file structure, executes a minimal simulation run, and verifies metrics like p-values and welfare fall within expected bounds. This prevents wasted resources by catching configuration errors early.

What metrics are checked during a multi-agent simulation sanity test?▼

A multi-agent simulation sanity test verifies that p-values, welfare, toxicity, and acceptance rate metrics fall within expected bounds, ensuring probabilities are between 0 and 1 and all values remain finite.

How do I check if my YAML scenario file is correctly formatted for multi-agent simulations?▼

To check if your YAML scenario file is correctly formatted, run a scenario validation script that parses the configuration, verifies required fields like agent types and topology, and confirms file existence in the scenarios directory.

Can I test for simulation crashes without running a full multi-hour epoch sweep?▼

Yes, you can test for simulation crashes without a full sweep by executing a short simulation run with a single seed for a few epochs. This minimal run validates basic metric calculation and identifies immediate runtime errors before committing computational resources.

What scenario configuration parameters are verified during a risk assessment sanity check?▼

During a risk assessment sanity check, verified parameters include agent types, governance, topology, and epoch or step counts. The validation ensures these fields exist and are structurally correct within the YAML configuration.

Why does my multi-agent simulation metric calculation return non-finite values?▼

Multi-agent simulation metric calculation returns non-finite values when scenario configurations contain invalid epoch or step counts, missing agent types, or incorrect topology definitions. Running a sanity check identifies these configuration issues before the full simulation begins.