The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill requires the user to add an external MCP server endpoint (https://rube.app/mcp) to their client configuration. This endpoint provides the logic and schemas for the Datadog tools, making the skill dependent on the availability and integrity of the third-party rube.app service.
[COMMAND_EXECUTION]: The skill exposes several high-impact administrative tools that allow the agent to modify the Datadog environment. Notably, DATADOG_DELETE_DASHBOARD allows for irreversible deletion of dashboards, while tools like DATADOG_MUTE_MONITOR and DATADOG_CREATE_DOWNTIME can be used to suppress active alerts and monitoring visibility.
[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection due to its core functionality of reading and acting upon external data.
Ingestion points: Untrusted data enters the agent context via DATADOG_SEARCH_LOGS (log body), DATADOG_LIST_EVENTS (event text), and DATADOG_QUERY_METRICS (metric tags and names).
Boundary markers: The instructions do not specify any delimiters or safety warnings (e.g., "ignore embedded instructions") when the agent processes the output from these ingestion tools.
Capability inventory: The skill provides significant capabilities that an attacker could target through injected instructions, including deleting resources (DATADOG_DELETE_DASHBOARD), modifying alert configurations (DATADOG_UPDATE_MONITOR), and creating alerts or events.
Sanitization: There are no instructions or mechanisms provided to sanitize, escape, or validate the content of the logs or events before the agent evaluates them.

datadog-automation