The Agent Skills Directory

[SAFE]: No malicious code, exfiltration patterns, or persistence mechanisms were detected. The skill contains instructional content and scaffolding scripts for machine learning operations and LLM integration consistent with industry standards.
[PROMPT_INJECTION]: The skill demonstrates patterns for Retrieval-Augmented Generation (RAG) and LLM application development that include vulnerability surfaces for indirect prompt injection.
Ingestion points: Untrusted data is ingested via variables like user_input in references/llm_integration_guide.md and {context} in references/rag_system_architecture.md.
Boundary markers: No delimiters or "ignore embedded instructions" warnings are present in the provided prompt templates.
Capability inventory: The skill defines provider classes for interacting with OpenAI and Anthropic APIs, alongside scripts for model deployment and RAG system building.
Sanitization: No input sanitization or filtering logic is included in the prompt interpolation examples.
[EXTERNAL_DOWNLOADS]: The skill references several well-known libraries and services (e.g., MLflow, Pinecone, Tecton, LangChain, OpenAI). It also includes a Dockerfile template that utilizes pip to install requirements. These are standard development practices for the intended use case.

senior-ml-engineer