The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill installs established Python libraries for RAG evaluation, including ragas, langchain-openai, datasets, and pandas from official registries.
[INDIRECT_PROMPT_INJECTION]: The skill processes data from RAG systems (questions, answers, and contexts) for evaluation by an LLM.
Ingestion points: The test_cases and Dataset structures in SKILL.md act as containers for untrusted external data.
Boundary markers: None identified in the provided code snippets.
Capability inventory: No high-risk capabilities like file system modification or shell execution are present.
Sanitization: No specific content filtering is implemented for the data being evaluated.

rag-evaluation