Compose a team (of people OR agents) by lifecycle posture, not job title, using the five product archetypes — Prototyper, Builder, Sweeper, Grower, Maintainer — and the rule that a healthy mix depends on product stage. Use when deciding which agents/people to deploy on a project, diagnosing why a team or agent bundle feels lopsided (all builders, no maintainers), choosing what posture to take on a given piece of work, staffing a pre-PMF vs growing vs mature product, or selecting an agent mix ...
Scanned 9/5/2026
Install to Claude Code
npx -y skills add The-AIOS/aios --skill team-archetypes --agent claude-codeInstalls into .claude/skills of the current project.
Are you the author of Team Archetypes?
Add the live security badge to your README — it updates automatically with every re-scan.
[](https://www.skillsdirectory.com/skills/the-aios-team-archetypes)More formats (shields.io, HTML) on the badges page.
---
name: team-archetypes
description: "Compose a team (of people OR agents) by lifecycle posture, not job title, using the five product archetypes — Prototyper, Builder, Sweeper, Grower, Maintainer — and the rule that a healthy mix depends on product stage. Use when deciding which agents/people to deploy on a project, diagnosing why a team or agent bundle feels lopsided (all builders, no maintainers), choosing what posture to take on a given piece of work, staffing a pre-PMF vs growing vs mature product, or selecting an agent mix in /7plan or /emerge. The team-composition complement to leverage-points: leverage-points says WHERE to intervene; this says WITH WHAT POSTURE."
---
# Team Archetypes — compose by posture, not by title
As engineering, product, design, and DS melt into a new kind of role, the durable unit of a team isn't the job function — it's the **posture** someone takes toward the work at a given stage. Boris Cherny (Claude Code) named five archetypes that recur across every function: the same person can be a Prototyper on one project and a Maintainer on another; a designer, an engineer, and a PM can each be any of the five. This skill is a lens for **composing a team — of humans OR of AIOS agents — by archetype, matched to product stage.**
## The five archetypes
| # | Archetype | What they do | Posture |
|---|---|---|---|
| 1 | **Prototyper** | Churns out brand-new ideas; most don't ship | Generative, throwaway-tolerant, speed over polish |
| 2 | **Builder** | Turns a prototype/idea into production-grade product/infra | Hardening, end-to-end ownership |
| 3 | **Sweeper** | Simplifies code/systems, unships, optimizes, cleans UI | Subtractive, entropy-fighting |
| 4 | **Grower** | Iterates a *built* product to improve Product-Market Fit | Experiment-driven, metric-chasing |
| 5 | **Maintainer** | Owns a mature system: secure, reliable, fast, efficient at scale | Stewardship, operational rigor |
Most people (and most good agents) span **2–3** of these. Archetype is **not** tied to job function — it's a disposition toward the work.
## The core move — match the mix to the stage
A healthy team needs a *mix*, and the right mix depends on where the product is:
| Product stage | Archetype mix it needs |
|---|---|
| **New / pre-PMF** | strong **1 + 2 + 3** (prototype, build, keep it simple) |
| **Growing / found PMF** | **2 + 3 + 4** + some **5** |
| **Strong PMF / scaling** | **3 + 4 + 5** + some **2** |
When you staff or compose for a project, **name the stage first, then assemble the matching mix.** A pre-PMF product staffed entirely with Maintainers will stall; a scaling product full of Prototypers will thrash.
## The reframe that prevents the category error
**Archetype is orthogonal to capability.** Capability = *what* someone/an agent can do (review code, threat-model, research a market). Archetype = *how/when* they deploy it, matched to stage. The two **multiply** — they don't collapse.
- Don't rename your roles/agents to the five archetypes. That fuses two independent axes and destroys the capability information (a "Prototyper" *of what*?).
- Instead, **tag** each role/agent with the archetype(s) its posture can serve (1-to-many), and let the *same* unit shift posture by stage.
## Applying it to AIOS agents
AIOS bundles agents by **domain** (sales / strategy / engineering / …). Archetype is the **orthogonal posture layer** on top. The engineering bundle maps roughly:
| Archetype | AIOS agent(s) | Posture-mode notes |
|---|---|---|
| Prototyper | `technical-cofounder` *(prototype mode)* | Throwaway-churn posture pre-PMF — explicitly *not* production-grade |
| Builder | `technical-cofounder`, `aios-builder` | Production-grade, end-to-end |
| Sweeper | `refactor-engineer` (+ `code-reviewer`, `/simplify` skill) | Simplify, unship, optimize — proactive entropy reduction on a mature codebase |
| Grower | `growth-engineer` | PMF-iteration on a *shipped* product — funnel, activation, retention, experiments |
| Maintainer | `security-engineer`, `bug-triager`, `code-documenter` | Reliability, security, docs-in-sync |
**Coverage read:** the engineering bundle now spans the full **build → grow → maintain** lifecycle. **Prototyper** is the one archetype with no dedicated agent — by design: it's a *posture-mode* of `technical-cofounder` (pre-PMF churn), not a missing capability. The rule it illustrates: fill a thin archetype with a posture-mode when the capability already exists, and a new agent only when a genuine capability is absent (as `growth-engineer` was) — never with a bundle restructure.
## Three use-sites
1. **Agent/team selection** — `/7plan` (weekly: name each project's stage, check the planned work matches the stage-appropriate mix) and `/emerge` (archetype-coverage gap → agent suggestion). *Not `/today`* — daily task-routing is tactical; stage→mix is a strategic concern.
2. **Posture self-location** (an agent reading this skill): a builder agent on a *pre-PMF* project takes the **Prototyper** posture (churn, throwaway, speed); on a *scaling* project, the **Builder/Maintainer** posture (hardening, rigor). Same agent, stage-appropriate posture.
3. **Project staging:** tag each project (Active / Idea) with a lifecycle stage so the mix-check has an input.
## Guardrails
- **Don't restructure bundles around archetypes.** Domain bundling works; archetype is a lens layered on top, not a replacement. Restructuring is churn against a taxonomy that already works.
- **Don't manufacture five agents per domain.** 5 archetypes × N domains is an explosion that re-buries the capability axis. Tag, don't multiply.
- **A thin archetype = a capability question, not a naming question.** "We have no Grower" means "we lack PMF-iteration capability," addressed via `/emerge` → a new *capability* agent — not by relabeling an existing one.
## Relationship to the rest of the system
Companion to `leverage-points` (where to intervene) and `sustainable-cadence` (at what pace). Worldview-neutral team science, like Meadows/Ostrom — it belongs to the invariant core, not a working-style template. Source: [Boris Cherny on product-role archetypes](references/cherny-archetypes-source.md).
Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.
No comments yet. Be the first to comment!