The Agent Skills Directory

[PROMPT_INJECTION]: The skill is designed to ingest and process untrusted data from external chat platforms, creating a surface for indirect prompt injection attacks.
Ingestion points: User messages are accessed via the message.text property within event handlers like onNewMention and onSubscribedMessage in SKILL.md.
Boundary markers: The provided implementation examples do not use delimiters or explicit instructions to the AI to ignore potentially malicious content embedded in user messages.
Capability inventory: The skill utilizes thread.post() to send messages, JSX-based interactive cards, and streaming AI responses back to external platforms.
Sanitization: There is no evidence of input sanitization or validation before the user-supplied text is passed to the thread.post() method or an AI agent's stream function.

chat-sdk