The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill exposes an attack surface for indirect prompt injection by reading untrusted data from external sources, specifically Discord messages and reactions (e.g., via DISCORDBOT_LIST_MESSAGES). This data is ingested into the agent's context and could contain malicious instructions. The skill also provides high-privilege capabilities, including sending messages, managing server roles (DISCORDBOT_CREATE_GUILD_ROLE), and executing webhooks (DISCORDBOT_EXECUTE_WEBHOOK). The instructions do not define boundary markers, delimiters, or sanitization protocols to isolate untrusted content from system instructions.\n- [EXTERNAL_DOWNLOADS]: The skill setup requires the user to configure a connection to an external Model Context Protocol (MCP) server at https://rube.app/mcp. This establishes a dependency on a third-party service outside of the whitelisted domains to provide the core functionality and tool logic for the skill.

discord-automation