The Agent Skills Directory

[SAFE]: The skill files consist solely of Markdown documentation and instructions. There are no scripts (.sh, .py, .js) or binaries included that could perform automated tasks.
[DATA_EXFILTRATION]: The skill does not contain any commands or functions capable of performing network requests or data transmission. No domains or external APIs are referenced for data output.
[INDIRECT_PROMPT_INJECTION]: The skill describes workflows for analyzing external data (e.g., call transcripts, survey responses in references/guidelines.md). However, because the skill lacks executable capabilities like shell access, file-writing, or network operations, there is no attack surface for exploiting malicious instructions embedded in that data.

content-strategy