The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes data from external sources (mentions and rosters) which could contain untrusted instructions.
Ingestion points: Untrusted data enters the agent context via the list_room_roster and list_mentions tools as described in SKILL.md.
Boundary markers: The instructions lack explicit boundary markers or directives for the agent to ignore instructions embedded within the mention text.
Capability inventory: The skill is explicitly configured as read-only and contains no file system, network, or command execution capabilities.
Sanitization: No sanitization or filtering of the retrieved Hilos data is performed before display.

hilos-room