{"id":77079,"plugin_id":"plugins_6ab405a6785081918aefad75d5faeef5","kind":"skill","collection_source":"plugin_package","comparison_source":null,"observed_at":"2026-11-03T00:12:22.797Z","digest":"24808a8c1b1fb4e76ac2ad11736322ff621c0bfdddf4baac65d3a220d82387d1","against":null,"payload":{"description":"Coordinate independent interview question design and review only when real subagents are available.","included_files":[],"name":"orchestrate-agents","skill_md_contents":"---\nname: orchestrate-agents\ndescription: Coordinate independent interview question design and review only when real subagents are available.\n---\n\n# Orchestrate Agents\n\n\nConsult `../../references/agent-coordination.md` and inspect actual host subagent capabilities. If genuine agents are available, the lead may delegate independent JD mapping and role-domain research, separately request answer rubric critique, then reconcile evidence. Do not spawn agents merely to show a panel; a live mock interview should have one coherent interviewer voice and one question per turn. Give agents nonoverlapping scopes and shared role/grounding constraints. Avoid transmitting full private resumes to unnecessary agents or services.\n\nIf no supported subagent execution exists, perform stages sequentially and never report 'multiple agents ran'. Each result must identify tasks, source evidence, verified actions, and unresolved issues. User consent and actual tool permissions govern retrieval or external access.\n"},"changes":[],"summary":"First saved snapshot. No earlier version is available for comparison.","summary_kind":"deterministic","summary_metadata":{}}