Create and upload evaluation datasets to LangSmith via CLI or SDK.
npx skills add https://github.com/lguibr/fh --skill langsmith-dataset-lguibr
Please help me install this Agent Skill. Skill: langsmith-dataset Source: https://github.com/lguibr/fh/tree/main/.agents/skills/langsmith-dataset Command: npx skills add https://github.com/lguibr/fh --skill langsmith-dataset-lguibr
LangSmith dataset workflows are often manual and error-prone; this skill automates creation, management, and uploading of evaluation datasets to LangSmith for testing and validation.
Create a LangSmith dataset named 'My Dataset' and upload it with examples using the CLI or SDK.
High-intent search queries and answers about installing and using this skill.
To create and upload LangSmith evaluation datasets, use the Python or TypeScript SDK to define dataset types, manage examples, and push data to a scoped project with authentication for reproducible testing.
LangSmith dataset types include final_response, single_step, trajectory, and RAG. Trajectory datasets capture multi-step agent actions, while RAG datasets focus on retrieval contexts and generated responses.
Yes, you can manage LangSmith datasets from the command line using CLI commands to create datasets, upload examples, and handle exports across Python and JavaScript/TypeScript environments.
Yes, authentication and project scoping are required to upload evaluation datasets to LangSmith, ensuring your testing data is securely attached to the correct project environment.
Collecting traces facilitates the end-to-end dataset lifecycle, allowing you to capture application runs and convert them into structured examples for uploading as reproducible LangSmith evaluation datasets.
anukkrit149Create and manage LangSmith evaluation datasets using the langsmith CLI.
Arize-aiBuild and run Phoenix evaluators for AI/LLM applications.
KappTech88Diagnose LLM behavior across tracing, evaluation, and monitoring workflows in LangSmith.
Diploma-pendingGenerate and upload evaluation datasets from LangSmith trace JSONL files.
