The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs users on how to use the node CLI to resolve the installation path of the @trigger.dev/sdk package. This is a standard development practice for locating local files and does not involve arbitrary command execution.
[EXTERNAL_DOWNLOADS]: The skill references the @trigger.dev/sdk Node.js package as a prerequisite. This is the official SDK provided by the vendor (Trigger.dev) and is an expected dependency for the described functionality.
[PROMPT_INJECTION]: The skill uses the keyword 'CRITICAL' to emphasize important architectural guidance regarding API usage. This is instructional language and not an attempt to override the AI agent's safety guardrails or instructions.
[PROMPT_INJECTION]: The skill describes an architecture for processing external data (session inputs and tool outputs), which constitutes an indirect prompt injection surface.
Ingestion points: Data enters the context via tool results (addToolOutput) and session channels (/realtime/v1/sessions/{id}/in/append).
Boundary markers: The provided documentation does not specify explicit delimiters or guardrails for sanitizing this input within the code examples.
Capability inventory: The skill utilizes fetch for communicating with Trigger.dev APIs and describes the use of tools for human-in-the-loop (HITL) interactions.
Sanitization: No explicit sanitization or validation routines for external message content are detailed in this reference guide.

trigger-chat-agent-advanced