theaiteam

A reusable, provider-agnostic AI team. Drop a project brief, get end-to-end delivery.

New here? Start at GETTING_STARTED.md — 10-minute onboarding from "what is this?" to "running it on my own project."

Roster (11 roles)

Role	File	Owns
Product Manager	team/roles/product-manager.md	Brief intake, PRDs, prioritization, customer research
System Architect	team/roles/architect.md	System design, tech-stack picks, API contracts, data models
Backend Engineer	team/roles/backend-engineer.md	Server code, DB, integrations, performance
Frontend Engineer	team/roles/frontend-engineer.md	Web UI, dashboards, client-side state
Mobile Engineer	team/roles/mobile-engineer.md	iOS (Swift) + Android (Kotlin) apps and SDKs
Designer	team/roles/designer.md	UX flows, design systems, mockups, brand visuals
QA Engineer	team/roles/qa-engineer.md	Test strategy, automation (Playwright), regression
Security Engineer	team/roles/security-engineer.md	OWASP review, threat models, static analysis
DevOps Engineer	team/roles/devops-engineer.md	CI/CD, infra, deploys, observability
Technical Writer	team/roles/technical-writer.md	API docs, READMEs, runbooks, changelogs
Marketing Strategist	team/roles/marketing-strategist.md	GTM, content, SEO, CRO, paid, analytics

Quickstart

# 1. Copy the brief template and fill it out
cp briefs/_template.md briefs/my-project.md
$EDITOR briefs/my-project.md

# 2. Sync team/ into your AI provider's expected location
./adapters/sync-all.sh

# 3. Open the project in Claude Code (or Codex / Gemini) and run:
#    /kickoff my-project

The orchestrator will:

1. Parse the brief and pick the right workflow
2. Create workspace/my-project/ with a task board
3. Dispatch roles sequentially or in parallel as the workflow demands
4. Track progress, handoffs, and PRs in the task board

Architecture

briefs/<project>.md  ─►  orchestrator  ─►  workflow  ─►  roles + skills  ─►  workspace/<project>/
                                                                              ├── plans/
                                                                              ├── artifacts/
                                                                              ├── task-board.md
                                                                              └── handoffs/

See docs/how-it-works.md for the deep dive and docs/plans/theaiteam-design.md for the design rationale.

Provider support

Provider	Adapter	Status
Claude Code	adapters/claude-code/	Primary
OpenAI Codex CLI	adapters/codex/	Supported
Gemini CLI	adapters/gemini/	Stub — extend as needed

Source of truth is plain markdown in team/. Adapters translate it to each provider's expected format.

License

Internal — Aman Singh.