Built an Open-Source Tool That Finds Missing Validation, Retries, and Error Handling in AI Agent Systems

We just released an open-source tool called **Trustabl Agent Analyzer** and we're looking for feedback from engineers building AI agents. The idea came from a problem we kept running into: agent systems often look good in demos, but production deployments reveal gaps in validation, retries, error handling, and tool configuration that are easy to miss during development. **What it does:** * Scans agent repositories built with Claude, OpenAI, Google ADK, MCP, and related frameworks * Analyzes agent, tool, and subagent definitions * Identifies missing validation rules, retry policies, and error handling patterns * Generates an instant report One thing we wanted from the start was privacy. The analyzer runs as a **local binary**, so you can point it at a local folder or GitHub repository and your code never leaves your machine. We're actively developing the project and would really appreciate honest feedback from people working with agent systems in production.