The Agent Skills Directory

[PROMPT_INJECTION]: The skill provides RAG (Retrieval-Augmented Generation) templates that interpolate external data from vector databases directly into LLM prompts without sanitization or boundary markers.
Ingestion points: External data enters the context via self.retrieve(question).passages in the RAG and SupportRAG classes (SKILL.md and examples.md).
Boundary markers: The templates (e.g., context, question -> answer) do not include delimiters or instructions to ignore embedded commands within the retrieved context.
Capability inventory: The ingested content influences LLM response generation via dspy.ChainOfThought (SKILL.md, examples.md). No high-privilege capabilities like shell execution are directly triggered by the RAG output in the provided examples.
Sanitization: There is no evidence of escaping, filtering, or validation of the retrieved passages before they are passed to the model.

dspy-qdrant