HN Showcase
Back to browse
GitHub Repository

Chaos engineering & reliability testing for multi-agent AI systems

3 starsPython

Find where multi-agent AI systems break before production

by surajkumar001·Jun 25, 2026·1 point·0 comments
Visit ProjectView on HN

AI Analysis

●MidNiche GemSolve My Problem

Zero-cost static analysis for agent topologies when live testing burns API credits.

Strengths
  • •No live LLM calls means instant feedback with zero API costs during development.
  • •Interactive D3 dashboard visualizes cascade failures and single points of failure clearly.
Weaknesses
  • •Static analysis can't catch runtime behaviors or emergent agent interactions.
  • •Narrow audience: only teams building multi-agent architectures, not single-agent apps.
Target Audience

AI engineers building multi-agent systems with CrewAI, LangGraph, or AutoGen

Similar To

Chaos Monkey · LangSmith · AgentOps

Similar Projects

Productivity●●Solid

I built a multi-agent Think Tank that calls out my bad decisions

Three persona-driven agents (Saul, Mike, Gus) feed a single 'Cook' synthesis layer that explicitly looks for contradictions and produces one actionable move — e.g., 'you say AI career but your logs show 90% bizdev.' The repo is primarily prompt templates and architecture docs, so it’s a practical playbook for anyone wanting to wire personal data into an opinionated multi-agent loop rather than a full platform.

Niche GemShip It
atulya_techtea
104mo ago