The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill requires the configuration of an external MCP server at https://rube.app/mcp. This is a functional requirement for accessing the Kit automation tools provided by the Rube/Composio ecosystem.
[DATA_EXFILTRATION]: The skill instructions facilitate the movement of subscriber data, including personally identifiable information (PII) like email addresses, between the local agent context and the external ConvertKit API via the Rube middleware. This data transfer is inherent to the skill's primary purpose of email marketing management.
[PROMPT_INJECTION]: The skill exhibits a surface for indirect prompt injection by ingesting untrusted data from the ConvertKit API (such as subscriber names, tags, and broadcast content).
Ingestion points: Untrusted data enters the agent context through the KIT_LIST_SUBSCRIBERS, KIT_LIST_BROADCASTS, and KIT_GET_BROADCAST tools.
Boundary markers: The instructions do not define specific delimiters or security warnings to prevent the agent from executing instructions potentially hidden within retrieved email content or subscriber metadata.
Capability inventory: The skill includes tools with side effects, including KIT_DELETE_SUBSCRIBER, KIT_DELETE_BROADCAST, and KIT_TAG_SUBSCRIBER, which could be targeted by injected instructions.
Sanitization: There is no evidence of input validation or content filtering for data retrieved from the remote API.

convertkit-automation