The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to subscribe to external event streams (GitHub Pull Requests, Slack channels) and inject those events as messages into the active AI session. This creates a surface for indirect prompt injection.
Ingestion points: External provider events are injected into the session from the @__relay_webhook__ identity.
Boundary markers: The instructions do not specify the use of delimiters or 'ignore embedded instructions' warnings for the injected content.
Capability inventory: The agent has capabilities to execute shell commands, mount remote filesystems, write files, and interact with external APIs (GitHub/Slack) through the Relay service.
Sanitization: There is no evidence of content sanitization, validation, or escaping of the inbound message payloads before they are processed by the agent.
[EXTERNAL_DOWNLOADS]: The skill requires the installation of the relayfile package via npm install -g relayfile. This is a tool provided by the vendor (AgentWorkforce) to facilitate the skill's primary purpose.
[COMMAND_EXECUTION]: The skill makes extensive use of shell commands and utility functions to interact with a local Unix socket and manage subscriptions. It uses tools like jq, curl, and a vendor-specific probe utility (agent-relay-probe) to communicate with the Agent Relay Desktop application.

subscribing-relay-webhooks-and-writebacks