The Agent Skills Directory

[Dynamic Execution] (MEDIUM): The skill implements dynamic module loading for both JavaScript and Python bot templates. Evidence in src/index.js: require(filePath) is used within a loop over local files. Evidence in main.py: bot.load_extension() is used to load modules based on directory listings. While standard for bot architectures, this facilitates execution of any code placed in these directories.
[Indirect Prompt Injection] (MEDIUM): The skill creates a vulnerability surface by processing untrusted data from external Discord users. 1. Ingestion points: interaction (JS) and ctx, user, and message (Python) parameters in src/commands/ping.js and main.py. 2. Boundary markers: Absent; user input is not delimited or flagged as untrusted. 3. Capability inventory: The examples demonstrate messaging (interaction.reply, ctx.respond) and local file discovery/loading. 4. Sanitization: Absent; the greet command in main.py directly interpolates the untrusted message string into the bot's response.

discord-bot-architect