The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides tools for spawning and managing worker agents, specifically through the add_agent tool and the agent-relay node agent spawn command. These tools allow the agent to define tasks that are executed by child processes or provider-backed workers.
[EXTERNAL_DOWNLOADS]: The skill references documentation and orchestration instructions hosted at agentrelay.com and links to related skill files in the AgentWorkforce GitHub repository. These are author-controlled resources used for guidance and coordination.
[PROMPT_INJECTION]: The skill's core functionality involves receiving and processing messages from other agents, which introduces an indirect prompt injection surface.
Ingestion points: Tools such as check_inbox, list_messages, and search_messages (SKILL.md) ingest content from external participants in the workspace.
Boundary markers: The instructions do not define specific delimiters or instructions to ignore commands embedded within the received messages.
Capability inventory: The agent has access to worker management tools (add_agent, remove_agent) and communication tools (send_dm, post_message) (SKILL.md).
Sanitization: There is no description of filtering or sanitizing the content of messages before the agent acts upon the information they contain.

using-agent-relay