AgentEval

Based on a GitHub repository with only 124 stars and a single personal tweet, AgentEval is in an early experimental stage without broad community adoption or third‑party validation; it should not be treated as a mature evaluation standard. It acts as the .NET counterpart to tools like RAGAS and DeepEval in the Python ecosystem, designed specifically for Microsoft Agent Framework and Microsoft.Extensions.AI. Developers can use it to automatically check whether an agent’s tool calls are correct, assess retrieval‑augmented generation quality, run stochastic evaluations and compare different models. In a demonstrated commerce‑agent scenario, it connects to an OpenClaw agent on a real shopping task and returns a verifiable score. No pricing or cost information is disclosed in the repository or socials. Given its open‑source nature, the toolkit itself is presumably free, but deployment requires a .NET environment and MAF, making it inaccessible to non‑.NET stacks. It suits .NET teams building MAF agents who need a quick internal evaluation pipeline; it is not suitable for Python‑based teams or those evaluating complex multi‑modal, long‑horizon agent behaviour.