The Agent Skills Directory

[SAFE]: The skill contains standard Python code templates for text processing, embedding generation, and retrieval evaluation using well-known libraries such as openai, sentence-transformers, numpy, scipy, nltk, and tiktoken. All code follows industry-standard patterns for building RAG (Retrieval-Augmented Generation) applications.
[INDIRECT_PROMPT_INJECTION]: The skill facilitates the ingestion and processing of external data through embedding and chunking pipelines, which is a common attack surface for indirect prompt injection.
Ingestion points: External text data is ingested via parameters in get_embeddings, chunk_by_tokens, and the DomainEmbeddingPipeline class.
Boundary markers: The templates do not implement specific boundary markers or "ignore" instructions for the ingested content, which is typical for embedding-only tasks.
Capability inventory: The skill includes text processing, chunking, local model inference via sentence-transformers, and network calls to the OpenAI API.
Sanitization: The skill includes a _default_preprocess function that performs basic sanitization by normalizing whitespace and removing non-standard special characters.

embedding-strategies