Team Builder

1---2name: team-builder3description: Interactive agent picker for composing and dispatching parallel teams4origin: community5---6 7# Team Builder8 9Interactive menu for browsing and composing agent teams on demand. Works with flat or domain-subdirectory agent collections.10 11## When to Use12 13- You have multiple agent personas (markdown files) and want to pick which ones to use for a task14- You want to compose an ad-hoc team from different domains (e.g., Security + SEO + Architecture)15- You want to browse what agents are available before deciding16 17## Prerequisites18 19Agent files must be markdown files containing a persona prompt (identity, rules, workflow, deliverables). The first `# Heading` is used as the agent name and the first paragraph as the description.20 21Both flat and subdirectory layouts are supported:22 23**Subdirectory layout** — domain is inferred from the folder name:24 25```26agents/27├── engineering/28│   ├── security-engineer.md29│   └── software-architect.md30├── marketing/31│   └── seo-specialist.md32└── sales/33    └── discovery-coach.md34```35 36**Flat layout** — domain inferred from shared filename prefixes. A prefix counts as a domain when 2+ files share it. Files with unique prefixes go to "General". Note: the algorithm splits at the first `-`, so multi-word domains (e.g., `product-management`) should use the subdirectory layout instead:37 38```39agents/40├── engineering-security-engineer.md41├── engineering-software-architect.md42├── marketing-seo-specialist.md43├── marketing-content-strategist.md44├── sales-discovery-coach.md45└── sales-outbound-strategist.md46```47 48## Configuration49 50Agents are discovered via two methods, merged and deduplicated by agent name:51 521. **`claude agents` command** (primary) — run `claude agents` to get all agents known to the CLI, including user agents, plugin agents (e.g. `everything-claude-code:architect`), and built-in agents. This automatically covers ECC marketplace installs without any path configuration.532. **File glob** (fallback, for reading agent content) — agent markdown files are read from:54   - `./agents/**/*.md` + `./agents/*.md` — project-local agents55   - `~/.claude/agents/**/*.md` + `~/.claude/agents/*.md` — global user agents56 57Earlier sources take precedence when names collide: user agents > plugin agents > built-in agents. A custom path can be used instead if the user specifies one.58 59## How It Works60 61### Step 1: Discover Available Agents62 63Run `claude agents` to get the full agent list. Parse each line:64- **Plugin agents** are prefixed with `plugin-name:` (e.g., `everything-claude-code:security-reviewer`). Use the part after `:` as the agent name and the plugin name as the domain.65- **User agents** have no prefix. Read the corresponding markdown file from `~/.claude/agents/` or `./agents/` to extract the name and description.66- **Built-in agents** (e.g., `Explore`, `Plan`) are skipped unless the user explicitly asks to include them.67 68For user agents loaded from markdown files:69- **Subdirectory layout:** extract the domain from the parent folder name70- **Flat layout:** collect all filename prefixes (text before the first `-`). A prefix qualifies as a domain only if it appears in 2 or more filenames (e.g., `engineering-security-engineer.md` and `engineering-software-architect.md` both start with `engineering` → Engineering domain). Files with unique prefixes (e.g., `code-reviewer.md`, `tdd-guide.md`) are grouped under "General"71- Extract the agent name from the first `# Heading`. If no heading is found, derive the name from the filename (strip `.md`, replace hyphens with spaces, title-case)72- Extract a one-line summary from the first paragraph after the heading73 74If no agents are found after running `claude agents` and probing file locations, inform the user: "No agents found. Run `claude agents` to verify your setup." Then stop.75 76### Step 2: Present Domain Menu77 78```79Available agent domains:801. Engineering — Software Architect, Security Engineer812. Marketing — SEO Specialist823. Sales — Discovery Coach, Outbound Strategist83 84Pick domains or name specific agents (e.g., "1,3" or "security + seo"):85```86 87- Skip domains with zero agents (empty directories)88- Show agent count per domain89 90### Step 3: Handle Selection91 92Accept flexible input:93- Numbers: "1,3" selects all agents from Engineering and Sales94- Names: "security + seo" fuzzy-matches against discovered agents95- "all from engineering" selects every agent in that domain96 97If more than 5 agents are selected, list them alphabetically and ask the user to narrow down: "You selected N agents (max 5). Pick which to keep, or say 'first 5' to use the first five alphabetically."98 99Confirm selection:100```101Selected: Security Engineer + SEO Specialist102What should they work on? (describe the task):103```104 105### Step 4: Spawn Agents in Parallel106 1071. Read each selected agent's markdown file1082. Prompt for the task description if not already provided1093. Spawn all agents in parallel using the Agent tool:110   - `subagent_type: "general-purpose"`111   - `prompt: "{agent file content}\n\nTask: {task description}"`112   - Each agent runs independently — no inter-agent communication needed1134. If an agent fails (error, timeout, or empty output), note the failure inline (e.g., "Security Engineer: failed — [reason]") and continue with results from agents that succeeded114 115### Step 5: Synthesize Results116 117Collect all outputs and present a unified report:118- Results grouped by agent119- Synthesis section highlighting:120  - Agreements across agents121  - Conflicts or tensions between recommendations122  - Recommended next steps123 124If only 1 agent was selected, skip synthesis and present the output directly.125 126## Rules127 128- **Dynamic discovery only.** Never hardcode agent lists. New files in the directory auto-appear in the menu.129- **Max 5 agents per team.** More than 5 produces diminishing returns and excessive token usage. Enforce at selection time.130- **Parallel dispatch.** All agents run simultaneously — use the Agent tool's parallel invocation pattern.131- **Parallel Agent calls, not TeamCreate.** This skill uses parallel Agent tool calls for independent work. TeamCreate (a Claude Code tool for multi-agent dialogue) is only needed when agents must debate or respond to each other.132 133## Examples134 135```136User: team builder137 138Claude:139Available agent domains:1401. Engineering (2) — Software Architect, Security Engineer1412. Marketing (1) — SEO Specialist1423. Sales (4) — Discovery Coach, Outbound Strategist, Proposal Strategist, Sales Engineer1434. Support (1) — Executive Summary144 145Pick domains or name specific agents:146 147User: security + seo148 149Claude:150Selected: Security Engineer + SEO Specialist151What should they work on?152 153User: Review my Next.js e-commerce site before launch154 155[Both agents spawn in parallel, each applying their specialty to the codebase]156 157Claude:158## Security Engineer Findings159- [findings...]160 161## SEO Specialist Findings162- [findings...]163 164## Synthesis165Both agents agree on: [...]166Tension: Security recommends CSP that blocks inline styles, SEO needs inline schema markup. Resolution: [...]167Next steps: [...]168```