The Agent Skills Directory

EXTERNAL_DOWNLOADS (MEDIUM): The skill requires the configuration of an external MCP server from an untrusted domain.
Evidence: Users are instructed to add https://rube.app/mcp as an MCP server. Since rube.app is not among the specified trusted organizations or repositories, the tool definitions and execution logic provided by this server are considered unverifiable dependencies.
PROMPT_INJECTION (LOW): The skill possesses a surface for indirect prompt injection because it reads and processes external Discord data.
Ingestion points: DISCORDBOT_LIST_MESSAGES and reaction listing tools (e.g., in SKILL.md) ingest data from potentially adversarial users.
Boundary markers: Absent; there are no instructions to use XML tags or other delimiters to isolate Discord message content from system instructions.
Capability inventory: The agent can send messages, manage guild roles (including ADMINISTRATOR level permissions), and execute webhooks, which could be abused if the agent is manipulated by content in Discord messages.
Sanitization: No sanitization or validation logic is defined for the external data ingested from Discord.

discord-automation