This repository contains ongoing research and a seed implementation for an AI-powered analytics analyst agent system. Everything here is a work in progress — structure, specs, and patterns are actively evolving.
- benchmarks/ — reference implementation and case studies: an agent management system with an evaluation harness, capability registry, and worked examples for an analytics analyst agent.
- seed/ — minimal starting point for new deployments: a bare-bones registry structure and CLAUDE.md spec to bootstrap a new agent from scratch.
Built collaboratively with Claude (Anthropic) as part of an exploration into how AI agents can support structured analytical work.