io.github.RudrenduPaul/agent-eval

Statistical regression testing for LLM agents: p-value, effect size, and CI on behavior change.

needs args probed 2026-10-14

MCP error -32000: Connection closed
usage: agent-regress [-h] [--version] {compare} ...
agent-regress: error: argument command: invalid choice: 'agent-regress-c

Run it: uvx agent-regress-cli==0.1.8 (stdio, probed in a container with the network off)
Host: pypi.org
Repository: https://github.com/RudrenduPaul/agent-eval
Registry: active, updated 2026-08-10 · registry entry
Reachable: 0 of the last 1 probe

Badge and feed

[![mcpcost](https://mcpcost.ulehla.dev/badge/io.github.RudrenduPaul/agent-eval.svg)](https://mcpcost.ulehla.dev/s/io.github.RudrenduPaul/agent-eval)

An entry for every probe that changed the result, the tool count or the token cost: Atom feed. Add it to a feed reader or a Slack RSS app to hear when the server stops answering or its tool set grows.