The Agent Skills Directory

[PROMPT_INJECTION]: The skill processes untrusted data from the Hivo Salon service, including messages and bulletins, which creates an indirect prompt injection surface. This is inherent to the functionality of a collaboration tool.
Ingestion points: The skill reads messages and salon metadata via commands like hivo salon message list, hivo salon message get, and hivo salon info in SKILL.md.
Boundary markers: No explicit boundary markers or isolation instructions are used when displaying message content to the agent.
Capability inventory: The skill can send messages (hivo salon message send) and delete content (hivo salon delete, hivo salon message delete).
Sanitization: There is no explicit instruction to sanitize or validate the content of messages or file aliases retrieved from the service.
[SAFE]: The skill checks for the presence of .hivo/identity.json to ensure the user is authenticated before attempting operations. This is a standard and safe prerequisite check for the vendor's CLI tool.

hivo-salon