The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests user-provided text through the summary_prompt_guide parameter, which is then used to influence the behavior of an AI model during the generation of subscription summaries. This represents an attack surface for indirect prompt injection where malicious instructions could be included in the prompt guide.
Ingestion points: summary_prompt_guide field in subscriptions-create and subscriptions-partial-update commands.
Boundary markers: None specified in the instructions.
Capability inventory: The skill has the ability to read insight and dashboard data and deliver content to external channels (Email, Slack, Webhooks).
Sanitization: No sanitization or validation of the prompt guide content is described.
[DATA_EXFILTRATION]: The skill allows the delivery of data snapshots to external destinations. While email and Slack integrations are controlled, the 'webhook' target_type allows users to specify an arbitrary URL in target_value. This functionality provides a path for data exfiltration if the agent is instructed to send sensitive dashboard data to an attacker-controlled endpoint.

managing-subscriptions