agent-teamlisted

ultrathink # Creating Agent Teams Launch coordinated agent teams for tasks requiring persistent cross-talk, shared state, and teammate lifecycle management. **Trigger:** `/agent-team <task-description>` or `/agent-team` ## Arguments - `$ARGUMENTS` — Free-text task description. If omitted, ask the user. - `--scenario <name>` — Use a pre-built template from [scenarios.md](scenarios.md). Skips Phase 2 (team design). Still requires user approval before launching. Valid: `advisor-guided-implementation`, `failure-investigation`, `research-analysis`, `codebase-migration`. - `--teammates N` — Override default teammate count. - `--all-opus` — Use Opus for all teammates (no advisor pattern). For tasks where Sonnet workers aren't sufficient (complex architecture, deep reasoning). - `--no-critic` — Skip critic review (read-only/exploratory tasks only). - `--fast` — Skip plan approval (urgent tasks only). ## When to Use ```dot digraph when_to_use { "Complex task needing\nmultiple workers?" [shape=diamond]; "Workers need to\ncommunicate findings?" [shape=diamond]; "Tasks fully\nindependent?" [shape=diamond]; "Single agent or subagent" [shape=box]; "dispatching-parallel-agents" [shape=box]; "creating-agent-teams" [shape=box style=filled fillcolor=lightgreen]; "Complex task needing\nmultiple workers?" -> "Single agent or subagent" [label="no"]; "Complex task needing\nmultiple workers?" -> "Workers need to\ncommunicate findings?" [label="yes"];