A Medium post introduces core/agent-eval, a Claude Code package the authors built to evaluate LLM agent workflows offline before production. The package reads an agent project, auto-detects what matters, and generates an evaluation suite.
Developer seeks practices for evaluating RAG-based agent pipelines in development
Developer seeks practices for evaluating RAG-based agent pipelines in development
Anthropic announced that Claude Code can now help users build evaluations and hillclimb on them. The announcement points to an article sharing guidance on eval design and Claude Code skills for improving applications. For practitioners, this extends Claude Code from coding assistance into automated eval-driven iteration on apps.
Anthropic announced that Claude Code can now help users build evaluations and hillclimb on them. The announcement points to an article sharing guidance on eval design and Claude Code skills for improving applications. For practitioners, this extends Claude Code from coding assistance into automated eval-driven iteration on apps.
Medium series publishes 'The Control Panel: Full-Stack Observability & Evaluation for Your Agent'
Medium series publishes 'The Control Panel: Full-Stack Observability & Evaluation for Your Agent'
A tutorial that builds a simple LLM agent and walks through evaluating it with Langfuse. It shows how to score agent behavior and interpret what the evaluation results reveal. Aimed at data scientists and developers wanting practical agent eval workflows.
A tutorial that builds a simple LLM agent and walks through evaluating it with Langfuse. It shows how to score agent behavior and interpret what the evaluation results reveal. Aimed at data scientists and developers wanting practical agent eval workflows.
Researchers introduce AgentCompass, an open-source, lightweight, and extensible infrastructure for evaluating LLM-based agents. It decouples evaluation into three independent components—Benchmark, Harness, and Environment—to improve reproducibility and reduce redundant engineering.
Researchers introduce AgentCompass, an open-source, lightweight, and extensible infrastructure for evaluating LLM-based agents. It decouples evaluation into three independent components—Benchmark, Harness, and Environment—to improve reproducibility and reduce redundant engineering.