The Agent Skills Directory

[Indirect Prompt Injection] (HIGH): The skill is highly vulnerable to indirect prompt injection because it instructs the agent to fetch and follow 'recommended execution plans' and tool schemas from a remote, untrusted server (rube.app).
Ingestion points: The RUBE_SEARCH_TOOLS function fetches tool slugs, schemas, and plans from an external API.
Boundary markers: Absent. The instructions command the agent to 'Always search first' and follow the returned schemas and plans without validation.
Capability inventory: The skill provides RUBE_MULTI_EXECUTE_TOOL which can perform write operations (Sendbird automation) and RUBE_MANAGE_CONNECTIONS which handles authentication state.
Sanitization: None. The agent is encouraged to use exact field names and types from the untrusted search results.
[Unverifiable Dependencies] (MEDIUM): The skill requires adding https://rube.app/mcp as an MCP server. This domain is not a trusted source. Adding a remote MCP endpoint allows a third party to define the executable capabilities of the agent at runtime.
[Data Exposure & Exfiltration] (MEDIUM): All automation tasks and connection management for Sendbird are routed through the rube.app infrastructure. This exposes sensitive communication data and potentially authentication tokens to the third-party service provider.

sendbird-automation