REPO PACKAGE
langchain-ai lca-skills
COMMUNITY2skills
0ready to use
LABSCO SUMMARY
langsmith-code-eval walks a nine-step collaborative process — it inspects your trace structure automatically, then builds category-based evaluation logic, and generates both the evaluator code and the experiment runner that calls it. langsmith-fetch is the read side of the same workflow: it pulls LangSmith traces so you can review conversation history and inspect tool calls while debugging agent behavior. Both need a LangSmith account and key; neither runs on anything less.
This is only useful if your team already traces its agents through LangSmith. The README says langchain-ai/lca-skills was "Built for LangChain Academy," LangChain's own training program, which puts the two skills closer to a worked teaching example of writing a LangSmith evaluator than to a general test-automation kit — if you don't trace through LangSmith at all, neither skill has anything for you.
READ THE FULL ANALYSIS
What is here is genuinely all there is. Two skills is the whole repository; there is no larger set behind it. The README's own Contributing section only invites pull requests adding more, and none had landed as of what we read.
WHAT'S INSIDE
2 showing · 2 total2 of 2
SETUP
Showing all 2 skills
langsmith-code-eval
Writes the code that grades your AI agent's answers automatically, so you get a score for each test case instead of reading through every run yourself.
langsmith-fetch
When an agent does something you did not expect, this fetches the saved record of that run out of LangSmith and writes it to your machine as files you can open and read.
No skill in this package matches that.
HOW TO GET IT
WHOLE PACKAGEInstall the whole package (2 skills)
npx skills add langchain-ai/lca-skillsSINGLE SKILLOr install a single skill
npx skills add langchain-ai/lca-skills --skill <name> --full-depthPick the skill name from the Skills tab — each entry there installs independently.