The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructions require the user to configure an external MCP server endpoint at https://rube.app/mcp. This configuration routes tool schemas and execution through a third-party domain not associated with the official vendor list or the skill author.
[DYNAMIC_EXECUTION]: The tool MIXPANEL_JQL_QUERY allows for the execution of arbitrary JavaScript Query Language (JQL) scripts within the Mixpanel environment. This presents a risk of dynamic code execution if the agent generates logic based on untrusted inputs.
[INDIRECT_PROMPT_INJECTION]: The skill demonstrates a significant attack surface by combining data ingestion from external analytics with high-privilege write and execution capabilities.
Ingestion points: Data retrieved from Mixpanel projects, including event names, property values, and user profiles via tools like MIXPANEL_QUERY_SEGMENTATION and MIXPANEL_QUERY_PROFILES (SKILL.md).
Boundary markers: The instructions do not define delimiters or specific "ignore" instructions to prevent the agent from obeying commands embedded in Mixpanel data.
Capability inventory: The skill possesses data mutation capabilities via MIXPANEL_PROFILE_BATCH_UPDATE and script execution via MIXPANEL_JQL_QUERY (SKILL.md).
Sanitization: There is no evidence of sanitization or strict schema validation for the data ingested from the Mixpanel API before it is processed by other tools.
[COMMAND_EXECUTION]: While not executing shell commands directly, the JQL interface (MIXPANEL_JQL_QUERY) functions as a specialized command execution environment for data processing logic.

mixpanel-automation