The Agent Skills Directory

[SAFE]: The skill is purely informational, serving as a documentation resource for developers. It does not contain any automated command execution, external network requests, or hardcoded credentials.
[INDIRECT_PROMPT_INJECTION]: The chat system implementation described in the guide handles untrusted text input from end-users which is later presented to administrators. This creates a vulnerability surface for indirect prompt injection if an AI agent is used to process or respond to messages within the admin dashboard.
Ingestion points: User-generated content is ingested via the support_messages data model and the WebSocket handlers (ChatChannel) defined in references/detailed-guide.md.
Boundary markers: The provided code snippets and architectural diagrams do not include instructions for using delimiters or boundary markers to isolate untrusted user content from agent instructions.
Capability inventory: The skill itself possesses no capabilities (such as file writes or shell execution); it provides templates that a user or agent would manually implement.
Sanitization: The guide does not specify sanitization or validation logic for the message content, focusing instead on the architectural flow and real-time synchronization.

chat-widget