Evals Skills

Name: Evals Skills
Availability: InStock
Author: jangletown

by jangletown·Mar 23, 2026·4 points·0 comments

Visit Project View on HN

AI Analysis

●●●BangerBig BrainSolve My Problem

Install eval pipelines via npm instead of reading docs, saving hours of setup.

Strengths

•Turns complex LLMOps onboarding into single-command installs for tracing and evals.
•Integrates directly with AI coding assistants for context-aware setup workflows.

Weaknesses

•Tied to LangWatch ecosystem, less useful if you use LangSmith.
•Requires adopting their specific skills framework instead of standard config files.

Post Description

Hello HN

I'm Rogerio, co-founder of LangWatch

This past month we've completely changed the way we onboard new customers now on LangWatch, instead of giving them instructions on how to instrument, cookbooks or UIs to build evals, or docs on how to write Scenario agent simulation tests, we simply give them skills now, or ready to copy-and-paste prompts.

This has reduced our onboarding time to only a few minutes, no more postponing evals because other priorities gets in the way.

We have now skills for everything for managing your agent lifecycle:

"Instrument my agent with open telemetry" "Write evaluations for my agent" "Write scenario tests and a CI pipeline for my agent" "Version my prompts"

and even more targeted recipes

"Check my agent doesn't give prescriptive advice" "Generate an evaluation dataset from my RAG knowledge base" "Test my CLI is well usable by other AI agents"

Check out more on LangWatch Skills directory above and lmk what you think