Assemble a Temporary Team

For the task below, select the most suitable members from the company's existing AI agents and assemble a temporary team to complete it together.

Task

$ARGUMENTS

Available Agents

These are all of the company's agents, defined in .claude/agents/:

Agent	File	Responsibilities
CEO	.claude/agents/ceo-bezos.md	Strategic decisions, business models, PR/FAQ, priorities
CTO	.claude/agents/cto-vogels.md	Technical architecture, technology selection, system design
Inversion Advisor	.claude/agents/critic-munger.md	Challenge decisions, identify fatal flaws, Pre-Mortem, prevent collective delusion
Product Design	.claude/agents/product-norman.md	Product definition, user experience, usability
UI Design	.claude/agents/ui-duarte.md	Visual design, design systems, color and typography
Interaction Design	.claude/agents/interaction-cooper.md	User flows, personas, interaction patterns
Full-Stack Development	.claude/agents/fullstack-dhh.md	Code implementation, technical solutions, development
QA	.claude/agents/qa-bach.md	Test strategy, quality control, bug analysis
DevOps/SRE	.claude/agents/devops-hightower.md	Deployment pipelines, CI/CD, infrastructure, monitoring and operations
Marketing	.claude/agents/marketing-godin.md	Positioning, brand, acquisition, content
Operations	.claude/agents/operations-pg.md	User operations, growth, community, PMF
Sales	.claude/agents/sales-ross.md	Sales funnels, conversion strategy
CFO	.claude/agents/cfo-campbell.md	Pricing strategy, financial models, cost control, unit economics
Research	.claude/agents/research-thompson.md	Market research, competitive analysis, industry trends, opportunity discovery

Execution Steps

1. Analyze the Task and Select Members

Select the 2-5 most relevant agents for the nature of the task. Selection principles:

• Select only those needed: More people is not always better; match the task's needs precisely
• Consider the collaboration chain: If the task spans design through development, ensure the key roles along that chain are represented
• Avoid redundancy: Do not select overlapping roles together

Briefly explain to the founder whom you selected and why, then immediately start assembling the team.

2. Create an Agent Team

Use the Agent Teams feature to assemble the temporary team:

• Create a team with a short, task-based team_name (English, kebab-case)
• Create a specific task for each member (TaskCreate), with enough context in the task description
• Use the Task tool to spawn each teammate with subagent_type set to general-purpose; inject the full content of the corresponding agent file into the prompt as its role definition
• When spawning a teammate, state in the prompt: its role definition, the task to complete, the cycle's runtime Language instruction, and that output documents belong under docs/<role>/

For frontend work, name one selected member as design owner without expanding the team just for a title. The owner and implementer must read .claude/skills/frontend-design.md. Pass the user task, agreed visual direction, key component decisions and acceptance states with the implementation handoff; the owner checks the implemented screens, not only a written design.

For a dedicated frontend finishing cycle, the design owner, implementer and verifier also use .claude/skills/frontend-polish/SKILL.md. Hand off the current product's screenshots and known functional behavior, the identity to retain, concrete defects to fix and the interface version already reviewed. Share only materials within the permitted project workspace. Complete and record the pass once for that version; later targeted reviews do not restart a full redesign.

3. Coordinate and Consolidate

• Coordinate the members' work as team lead
• Collect their outputs and consolidate them into a unified conclusion or proposal
• If there are disagreements, present each position for the founder to decide
• Clean up team resources after completion

Notes

• Follow the runtime product Language instruction for user-facing communication and documents throughout the current product, including pauses and restarts. A requested language change applies to the next product. If no product language is assigned, follow explicit human language instructions or the user's language. Keep technical terms in English. The language of this skill's source does not set the output language
• Leave language preference changes to the human-operated settings interface; do not edit saved language configuration or the current product's language state
• Store each member's output documents under docs/<role>/ as agreed
• The team is temporary and dissolves when the task is complete
• The founder is the final decision-maker; agents provide advice but do not replace that decision-making authority