The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process potentially untrusted data from external sources which could be used to influence the agent's behavior.
Ingestion points: The agent retrieves data from Datadog using tools like DATADOG_SEARCH_LOGS (log messages), DATADOG_LIST_EVENTS (event text), and DATADOG_QUERY_METRICS (metric metadata), all of which can contain content controlled by external actors or systems.
Boundary markers: The skill instructions do not specify any delimiters or instructions for the agent to ignore potentially malicious commands embedded within the retrieved Datadog data.
Capability inventory: The skill provides the agent with significant operational capabilities, including the ability to delete dashboards (DATADOG_DELETE_DASHBOARD), modify monitoring configurations (DATADOG_CREATE_MONITOR, DATADOG_UPDATE_MONITOR), and schedule maintenance windows (DATADOG_CREATE_DOWNTIME), which could be abused if an injection is successful.
Sanitization: There are no instructions or automated mechanisms described for sanitizing or validating the data fetched from Datadog before the agent acts upon it.

datadog-automation