team

name	team
description	Multi-agent team coordinator that breaks complex tasks into parallel or sequential workstreams, delegates to specialized skills, and merges results.
layer	orchestrator
category	orchestration
triggers	["/team","coordinate a team","parallelize this work","split this into workstreams","multi-agent build"]
inputs	["Complex task description requiring multiple concurrent or sequential workstreams","Optional team composition preferences","Optional priority ordering","Optional dependency graph between subtasks"]
outputs	["Coordinated results from all workstreams","Dependency resolution report","Merged final output with conflict resolution","Team execution summary with per-workstream status"]
linksTo	["cook","plan","scout","research","brainstorm","code-review","test","fix","refactor","optimize","debug","kanban","sequential-thinking","mermaid","docs-writer"]
linkedFrom	["kanban"]
preferredNextSkills	["ship","code-review","pr-writer"]
fallbackSkills	["cook","plan"]
riskLevel	high
memoryReadPolicy	full
memoryWritePolicy	always
sideEffects	["Creates or modifies multiple source files across workstreams","Creates git commits per workstream or as a batch","May create temporary coordination artifacts","Modifies project state across multiple modules"]

Team

Purpose

Team is the multi-agent coordinator for tasks too large or complex for a single skill to handle alone. It decomposes a big objective into workstreams, identifies dependencies between them, assigns each workstream to the appropriate skill (often cook for feature work, but also research, refactor, audit, etc.), manages execution order, and merges results.

Team thinks in terms of a dependency graph: which tasks can run in parallel, which must be sequential, and where integration points exist. It acts as the project manager that keeps all workstreams aligned and resolves conflicts when their outputs overlap.

Workflow

Phase 1: Decompose

1. Parse the objective -- Understand the full scope of the complex task.
2. Invoke plan -- Create a high-level project plan that identifies discrete workstreams.
3. Invoke sequential-thinking -- For tasks with non-obvious dependencies, reason through the execution order.
4. Build the dependency graph -- Map out which workstreams depend on which, using mermaid to visualize if the graph is complex.

Phase 2: Assign

5. Map workstreams to skills -- Each workstream gets a primary skill:
   • Feature work -> cook
   • Research tasks -> research
   • Refactoring -> refactor
   • Performance work -> optimize
   • Exploration -> brainstorm
   • Testing -> test
6. Set execution order -- Determine which workstreams can run in parallel and which must wait for dependencies.
7. Present the team plan -- Show the user the workstream breakdown, assignments, and execution order. Get confirmation for high-risk decompositions.

Phase 3: Execute

8. Run independent workstreams -- Start all workstreams with no unmet dependencies.
9. Monitor progress -- Track completion status of each workstream.
10. Resolve blockers -- If a workstream fails or gets stuck:
    • Invoke debug or fix for code issues
    • Invoke for design ambiguity