The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill implements patterns for ingesting untrusted data from local directories and external APIs for use in LLM context windows.
Ingestion points: Data is ingested via SimpleDirectoryReader (local files) and ApiDocumentReader (remote API calls via fetch) in examples/ingestion.md and examples/core.md.
Boundary markers: The examples do not demonstrate the use of explicit delimiters or instructions to ignore embedded commands within the retrieved text chunks.
Capability inventory: The skill enables agents to perform semantic search, tool calling, and multi-agent orchestration. While it advises against unsafe practices (e.g., "never eval()"), the synthesis of responses from retrieved data could be influenced by malicious instructions embedded in the source documents.
Sanitization: There is no evidence of sanitization, filtering, or validation of the content fetched from the ApiDocumentReader before it is converted into Document objects and indexed.

ai-orchestration-llamaindex