The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process real-time message data and history, which constitutes an indirect prompt injection surface.
Ingestion points: Data is retrieved from external sources via tools such as get_pubnub_messages and subscribe_and_receive_pubnub_messages as described in SKILL.md and references/incident-runbook.md.
Boundary markers: While the skill emphasizes structured logging and schema versioning, it does not explicitly define delimiters to prevent the agent from interpreting instructions embedded within the message payloads.
Capability inventory: The skill has the capability to perform network operations (send_pubnub_message) and read historical data based on the content of processed messages.
Sanitization: No specific sanitization or filtering logic for the raw message content is provided to mitigate the risk of instructions embedded in the data.

pubnub-observability