The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill provides numerous examples of bots that ingest untrusted data from external sources (Discord messages and interactions).
Ingestion points: Untrusted data enters the agent context through on_message(message) events in SKILL.md and examples.md, and through discord.Interaction objects in slash command handlers.
Boundary markers: The provided examples do not implement specific delimiters or 'ignore' instructions for user-provided strings before processing.
Capability inventory: The skill includes capabilities to perform server management actions like member.kick(), member.ban(), and channel.purge() in examples.md and reference.md.
Sanitization: No specific sanitization or validation of the content from message.content or interaction parameters is demonstrated before they are used in command responses or moderation actions.
[DYNAMIC_EXECUTION]: The skill demonstrates the standard library pattern for modular bot design using dynamic loading.
Evidence: Examples in examples.md and reference.md show the use of bot.load_extension("cogs.moderation") to load bot modules at runtime. This is the idiomatic way to organize discord.py bots and uses static string paths in the examples.

discord-py