The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill facilitates a Retrieval-Augmented Generation (RAG) workflow using the antfly agents retrieval command, which is susceptible to indirect prompt injection.
Ingestion points: Data is loaded into the system using antfly load from external source files (e.g., wiki-articles.json as mentioned in SKILL.md).
Boundary markers: The provided instructions do not specify the use of delimiters or instructions to the LLM to ignore embedded commands within the retrieved documents.
Capability inventory: The skill uses the antfly CLI for searching, reranking, and generating answers via local or remote LLM providers (Ollama).
Sanitization: There is no evidence of sanitization or content filtering of the retrieved data before it is interpolated into the generation prompt.

antfly-cli