The Agent Skills Directory

[PROMPT_INJECTION]: The skill processes untrusted user input from chatbot platforms, creating an indirect prompt injection surface.\n
Ingestion points: The ctx.body property is accessed in SKILL.md and patterns.md to process incoming messages.\n
Boundary markers: No specific delimiters or "ignore" instructions are used to wrap user input in the provided code patterns.\n
Capability inventory: The bot can perform network requests (fetch in patterns.md), file operations (provider.saveFile in patterns.md), and send outgoing messages (bot.sendMessage in patterns.md).\n
Sanitization: Basic validation logic (e.g., regex for emails) is demonstrated in patterns.md, but there is no universal sanitization of the ctx.body input.\n- [COMMAND_EXECUTION]: The documentation includes deployment instructions that recommend granting high-privilege capabilities to the bot container.\n
Evidence: providers.md contains a Docker command using the --cap-add SYS_ADMIN flag.\n
Context: This privilege is noted as necessary for the Baileys provider which relies on Chromium-based automation.\n- [DATA_EXFILTRATION]: The skill provides patterns for the bot to transmit data to external servers via network operations.\n
Evidence: patterns.md demonstrates using the fetch API to send user data (such as phone numbers) to an external API endpoint (https://my.app/api/check).

builderbot-code-skill