Show HN: Better Agents CLI (github.com)

LangWatch released Better Agents, a CLI scaffolding tool and standards suite that jumpstarts production-ready agent projects by enforcing a common structure, testing, evaluation, observability and prompt versioning. The tool generates an opinionated repo (app/, tests/scenarios/, tests/evaluations/, prompts/, prompts.json, AGENTS.md, .mcp.json, etc.), includes scenario-based end-to-end tests and Jupyter evaluation notebooks, and ships with guidance (Agent Testing Pyramid) so features, prompts and behaviors are systematically tested and measurable. Prompts are stored in YAML and registered via prompts.json for team collaboration, while .mcp.json preconfigures MCPs so your coding assistant (Kilocode, Claude Code, Cursor, etc.) becomes fluent in your chosen agent framework (Agno, Mastra, etc.). For the AI/ML community this matters because it standardizes agent development and CI-friendly verification: reproducible prompt versioning, built-in evaluations for RAG/classification subtasks, and observability hooks reduce drift and speed iterations from prototype to production. The CLI walks you through language, framework, LLM provider and API key setup; it requires Node.js 22+ and npm/pnpm and is available via npm or npx. MIT-licensed and framework-agnostic, Better Agents aims to make multi-developer agent projects maintainable, auditable and easier to evaluate at scale.