Show HN: Eval-skills for automatic agent improvement

Category: ai-ml

Tags: ai-evaluation, agent-skills, llm-ops, developer-tools, workflow

Score: 6.3/10 (Innovation: 6, Technical: 5, Documentation: 8, Utility: 6)

Eval Skills provides a set of coding-agent skills that guide developers through a structured workflow for evaluating and improving AI applications, from inspecting real executions to building validated graders and running controlled experiments. It stands out by offering a stage-based, evidence-driven approach that integrates with existing tools and avoids metric catalogs, making evaluation more practical and actionable for product teams.

Target audience: AI engineers, product managers, and domain experts building and maintaining LLM-based applications

Repository: https://github.com/confident-ai/eval-skills · Python · Apache-2.0 · 11 stars

View on Hacker News