The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill facilitates the ingestion of untrusted data from external Discord environments, creating a potential surface for indirect prompt injection attacks.
Ingestion points: The agent can read content from external sources via tools such as DISCORDBOT_LIST_MESSAGES, DISCORDBOT_GET_GUILD_MEMBER, and DISCORDBOT_LIST_MESSAGE_REACTIONS_BY_EMOJI as described in SKILL.md.
Boundary markers: There are no instructions provided to use delimiters or specific warnings to ignore commands or instructions that may be embedded within the Discord messages or profile data being processed.
Capability inventory: The skill grants the agent extensive write and administrative capabilities, including DISCORDBOT_CREATE_MESSAGE, DISCORDBOT_CREATE_GUILD_ROLE, DISCORDBOT_ADD_GUILD_MEMBER_ROLE, DISCORDBOT_DELETE_GUILD_ROLE, and DISCORDBOT_EXECUTE_WEBHOOK.
Sanitization: The instructions do not include requirements for sanitizing, escaping, or validating external content before the agent uses it to perform downstream operations or decision-making.
[EXTERNAL_DOWNLOADS]: The skill configuration requires connecting to a remote MCP endpoint.
The setup section in SKILL.md directs users to add https://rube.app/mcp as an MCP server. This establishes a communication channel with an external service provider to fetch tool schemas and execute operations.

discord-automation