Team Brainstorm

Orchestrate multi-agent brainstorming: generate ideas -> challenge assumptions -> synthesize -> evaluate. Supports Quick, Deep, and Full pipelines with Generator-Critic loop.

Architecture

Skill(skill="team-brainstorm", args="topic description")
                    |
         SKILL.md (this file) = Router
                    |
     +--------------+--------------+
     |                             |
  no --role flag              --role <name>
     |                             |
  Coordinator                  Worker
  roles/coordinator/role.md    roles/<name>/role.md
     |
     +-- analyze -> dispatch -> spawn workers -> STOP
                                    |
                    +-------+-------+-------+
                    v       v       v       v
                 [ideator][challenger][synthesizer][evaluator]

Role Registry

| Role | Path | Prefix | Inner Loop | |------|------|--------|------------| | coordinator | roles/coordinator/role.md | — | — | | ideator | roles/ideator/role.md | IDEA- | false | | challenger | [roles/challenger/role.md](roles/challenger/role.md) | CHALLENGE- | false | | synthesizer | roles/synthesizer/role.md | SYNTH- | false | | evaluator | [roles/evaluator/role.md](roles/evaluator/role.md) | EVAL- | false |

Role Router

Parse $ARGUMENTS:

• Has --role <name> -> Read roles/<name>/role.md, execute Phase 2-4
• No --role -> roles/coordinator/role.md, execute entry router

Delegation Lock

Coordinator is a PURE ORCHESTRATOR. It coordinates, it does NOT do.

Before calling ANY tool, apply this check:

| Tool Call | Verdict | Reason | |-----------|---------|--------| | spawn_agent, wait_agent, close_agent, send_message, followup_task | ALLOWED | Orchestration | | list_agents | ALLOWED | Agent health check | | request_user_input | ALLOWED | User interaction | | mcp__ccw-tools__team_msg | ALLOWED | Message bus | | Read/Write on .workflow/.team/ files | ALLOWED | Session state | | Read on roles/, commands/, specs/ | ALLOWED | Loading own instructions | | Read/Grep/Glob on project source code | BLOCKED | Delegate to worker | | Edit on any file outside .workflow/ | BLOCKED | Delegate to worker | | Bash("ccw cli ...") | BLOCKED | Only workers call CLI | | Bash running build/test/lint commands | BLOCKED | Delegate to worker |

If a tool call is BLOCKED: STOP. Create a task, spawn a worker.

No exceptions for "simple" tasks. Even a single-file read-and-report MUST go through spawn_agent.

Shared Constants

• Session prefix: BRS
• Session path: .workflow/.team/BRS-<date>-<slug>/
• CLI tools: ccw cli --mode analysis (read-only), ccw cli --mode write (modifications)
• Message bus: mcp__ccw-tools__team_msg(session_id=<session-id>, ...)

Worker Spawn Template

Coordinator spawns workers using this template:

spawn_agent({
  agent_type: "team_worker",
  task_name: "<task-id>",
  fork_turns: "none",
  message: `## Role Assignment
role: <role>
role_spec: <skill_root>/roles/<role>/role.md
session: <session-folder>
session_id: <session-id>
requirement: <topic-description>
inner_loop: false

Read role_spec file (<skill_root>/roles/<role>/role.md) to load Phase 2-4 domain instructions.

## Task Context
task_id: <task-id>
title: <task-title>
description: <topic-description>
pipeline_phase: <pipeline-phase>

## Upstream Context
<prev_context>`
})

After spawning, use wait_agent({ timeout_ms: 1800000 }) to collect results. If result.timed_out, send STATUS_CHECK via followup_task (wait 3 min), then FINALIZE with interrupt (wait 3 min), then mark timed_out and close agents. Use close_agent({ target: <name> }) each worker.

Parallel ideator spawn (Full pipeline with N angles):

When Full pipeline has N parallel IDEA tasks, spawn N distinct team-worker agents named ideator-1, ideator-2, etc.

spawn_agent({
  agent_type: "team_worker",
  task_name: "ideator-<N>",
  fork_turns: "none",
  message: `## Role Assignment
role: ideator
role_spec: <skill_root>/roles/ideator/role.md
session: <session-folder>
session_id: <session-id>
requirement: <topic-description>
agent_name: ideator-<N>
inner_loop: false

Read role_spec file (<skill_root>/roles/ideator/role.md) to load Phase 2-4 domain instructions.

## Task Context
task_id: <task-id>
title: <task-title>
description: <topic-description>
pipeline_phase: <pipeline-phase>

## Upstream Context
<prev_context>`
})

After spawning, use wait_agent({ timeout_ms: 1800000 }) to collect results. If result.timed_out, send STATUS_CHECK via followup_task (wait 3 min), then FINALIZE with interrupt (wait 3 min), then mark timed_out and close agents. Use close_agent({ target: <name> }) each worker.

Model Selection Guide

| Role | model | reasoning_effort | Rationale | |------|-------|-------------------|-----------| | Ideator (IDEA-) | (default) | high | Creative divergent thinking requires full reasoning | | Challenger (CHALLENGE-) | (default) | high | Critical analysis of ideas needs deep reasoning | | Synthesizer (SYNTH-) | (default) | medium | Aggregation and convergence over generation | | Evaluator (EVAL-) | (default) | medium | Scoring and ranking against criteria |

Override model/reasoning_effort in spawn_agent when cost optimization is needed:

spawn_agent({
  agent_type: "team_worker",
  task_name: "<task-id>",
  fork_turns: "none",
  model: "<model-override>",
  reasoning_effort: "<effort-level>",
  message: "..."
})

User Commands

| Command | Action | |---------|--------| | check / status | View execution status graph, no advancement | | resume / continue | Check worker states, advance next step |

Session Directory

.workflow/.team/BRS-<date>-<slug>/
├── session.json                    # Session metadata + pipeline + gc_round
├── task-analysis.json              # Coordinator analyze output
├── .msg/
│   ├── messages.jsonl              # Message bus log
│   └── meta.json                   # Session state + cross-role state
├── wisdom/                         # Cross-task knowledge
│   ├── learnings.md
│   ├── decisions.md
│   ├── conventions.md
│   └── issues.md
├── ideas/                          # Ideator output
│   ├── idea-001.md
│   └── idea-002.md
├── critiques/                      # Challenger output
│   ├── critique-001.md
│   └── critique-002.md
├── synthesis/                      # Synthesizer output
│   └── synthesis-001.md
└── evaluation/                     # Evaluator output
    └── evaluation-001.md

Specs Reference

• specs/pipelines.md — Pipeline definitions and task registry

v4 Agent Coordination

Message Semantics

| Intent | API | Example | |--------|-----|---------| | Share idea context across ideators | send_message | Send seed ideas to running ideator for cross-pollination | | Not used in this skill | followup_task | No resident agents -- all workers are one-shot | | Check running agents | list_agents | Verify parallel ideator health during Full pipeline |

Parallel Ideator Coordination (Full Pipeline)

Full pipeline spawns N parallel ideators for different brainstorming angles. Use batch spawn + wait:

// Spawn N ideators in parallel, each with a different angle
const ideatorNames = ["IDEA-001", "IDEA-002", "IDEA-003"]
for (const name of ideatorNames) {
  spawn_agent({ agent_type: "team_worker", task_name: name, fork_turns: "none", ... })
}
wait_agent({ timeout_ms: 1800000 })  // 30 min — apply timeout cascade if timed_out
// Collect all idea outputs, feed to challenger as upstream context

Agent Health Check

Use list_agents({}) in handleResume and handleComplete:

// Reconcile session state with actual running agents
const running = list_agents({})
// Compare with tasks.json active_agents
// Reset orphaned tasks (in_progress but agent gone) to pending

Named Agent Targeting

Workers are spawned with task_name: "<task-id>" enabling direct addressing:

• send_message({ target: "IDEA-002", message: "..." }) -- share context from another ideator
• close_agent({ target: "IDEA-001" }) -- cleanup by name after wait_agent returns

Error Handling

| Scenario | Resolution | |----------|------------| | Unknown command | Error with available command list | | Role not found | Error with role registry | | CLI tool fails | Worker fallback to direct implementation | | Fast-advance conflict | Coordinator reconciles on next callback | | Completion action fails | Default to Keep Active | | Generator-Critic loop exceeds 2 rounds | Force convergence to synthesizer | | No ideas generated | Coordinator prompts with seed questions |