The Agent Skills Directory

Local Data Processing: The skill reads from local directories (case/ and policies/) and writes to an output/ directory. This behavior is consistent with its stated purpose of medical document processing and administrative task automation.
Command Usage (ripgrep): The workflow utilizes rg (ripgrep) for searching local policy documents. This is an efficient and standard way to locate information within a file system and does not involve risky shell operations or external script execution.
Indirect Prompt Injection Surface: Because the skill processes external text data such as transcripts and policy documents, it has a surface for indirect prompt injection. Instructions embedded within these source files could theoretically influence the content of the generated output/ files. The potential impact is limited as the skill does not have network access or administrative privileges.

prior-auth-packet-builder