One prompt. A PM coordinator plus 8 specialist agents. A fully built application.
Give Claude Code a business objective, and The Fabric coordinates a project manager role plus 8 specialist agents to deliver a working application — with complete documentation, tests, security review, and deployment config.
/fabric Build me a task management API with user auth and team workspaces
The workflow is designed to run end-to-end from that initial objective.
Business Objective
│
▼
┌────────────────────────┐
│ fabric-pm │
│ Project Manager │
│ (orchestrator) │
└───────────┬────────────┘
│
┌─────────────────┼─────────────────┐
▼ ▼ ▼
┌─────────────┐ ┌─────────────┐ ...
│ Phase 1 │ │ Phase 2 │
│ fabric-ba │ │ (parallel) │
│ Business │ │ ┌────────┐ │
│ Analyst │ │ │architect│ │
└──────┬──────┘ │ ├────────┤ │
│ │ │ ux │ │
▼ │ └────────┘ │
Requirements └──────┬──────┘
& User Stories │
▼
Architecture, API
Contracts, UI Specs
│
┌──────────────┤
▼ ▼
┌─────────────┐ ┌─────────────┐
│ Phase 3 │ │ Phase 4 │
│ fabric- │ │ (parallel) │
│ dev-lead │ │ ┌────────┐ │
│ + parallel │ │ │ qa │ │
│ sub-agents │ │ ├────────┤ │
└──────┬──────┘ │ │security│ │
│ │ └────────┘ │
▼ └──────┬──────┘
Working Code │
▼
Tests + Security
Review + Fixes
(up to 3 iterations)
│
┌──────────────┤
▼ ▼
┌─────────────┐ ┌─────────────┐
│ Phase 5 │ │ Phase 6 │
│ fabric- │ │ fabric- │
│ devops │ │ docs │
└──────┬──────┘ └──────┬──────┘
│ │
▼ ▼
Docker, CI/CD README, User Guide
Deploy Config API Docs
│
▼
┌────────────────────┐
│ DELIVERED │
│ Working app + │
│ 18 doc files + │
│ Full test suite │
└────────────────────┘
| Phase | Agent(s) | What happens | Parallel? |
|---|---|---|---|
| 0 | Project Manager | Charter, workspace setup, team/task creation | |
| 1 | Business Analyst | Requirements + user stories with acceptance criteria | |
| 2 | Architect + UX Engineer | System design, API contracts, wireframes, design system | Yes |
| 3 | Lead Developer | Full implementation with parallel sub-agents per module | |
| 4 | QA + Security Engineer | Unit/integration/e2e tests (Playwright), STRIDE threat model, OWASP review | Yes |
| 4-fix | Lead Developer | Fix critical/high issues, re-test (up to 3 iterations) | |
| 5 | DevOps Engineer | Dockerfile, CI/CD, deployment docs | |
| 6 | Technical Writer | README, user guide, API docs | |
| 7 | Project Manager | Final verification, cleanup, and delivery summary |
The QA agent produces a comprehensive test suite:
| Test Type | Directory | What's tested |
|---|---|---|
| Unit | tests/unit/ |
Core business logic |
| Integration | tests/integration/ |
API endpoints |
| E2E | tests/e2e/ |
Critical user flows via Playwright |
| Frontend | tests/frontend/ |
UI components, responsive behavior, form validation |
| Accessibility | Included in e2e | WCAG checks via @axe-core/playwright |
# Add the marketplace
claude plugin marketplace add mac8005/fabric-sdlc
# Install the plugin
claude plugin install fabric-sdlc- Claude Code with Agent Teams enabled (
CLAUDE_CODE_EXPERIMENTAL_AGENT_TEAMS) - Recommended: Claude Opus for best orchestration results
The plugin bundles 12 permissively licensed design and developer skills. Most are wired into the SDLC phases — the PM instructs the relevant agent to load them — and two are general-purpose extras.
| Skill | What it does | SDLC phase |
|---|---|---|
frontend-design |
Distinctive, non-generic UI design | 2, 3 |
theme-factory |
10 pre-built professional themes | 2 |
brand-guidelines |
Apply brand colors & typography | 2 |
canvas-design |
PNG/PDF diagrams, mockups, visuals | 2 |
algorithmic-art |
Generative interactive visuals (p5.js) | 2 |
web-artifacts-builder |
Complex React / shadcn/ui artifacts | 3 |
claude-api |
Anthropic API / Agent SDK integration | 3 |
mcp-builder |
Build MCP servers | 3 |
webapp-testing |
Playwright e2e & UI testing | 4 |
internal-comms |
Status reports & delivery summaries | 6, 7 |
skill-creator |
Author and optimize Claude skills | — |
slack-gif-creator |
Slack-optimized animated GIFs | — |
Document export skills for PDF, Word, PowerPoint, and spreadsheets are not bundled (licensing). If you have compatible export skills installed separately, Fabric uses them as optional post-processing steps.
# With a business objective
/fabric Build me an invoice API with PDF export and email notifications
# Interactive mode — asks what to build
/fabricAfter execution, your directory contains a complete project:
./
├── docs/ # 18 core docs (+ 07-tech-debt.md if needed)
│ ├── 00-project-charter.md # Business objective and scope
│ ├── 00-progress-log.md # Full audit trail of all phases
│ ├── 01-requirements.md # Requirements with IDs (REQ-001...)
│ ├── 01-user-stories.md # User stories with Given/When/Then
│ ├── 02-architecture.md # System design + ADRs
│ ├── 02-tech-stack.md # Technology selections + justification
│ ├── 02-api-contracts.md # API specs with examples
│ ├── 02b-wireframes.md # ASCII wireframes
│ ├── 02b-ui-spec.md # UI component specs
│ ├── 02b-design-system.md # Colors, typography, spacing
│ ├── 03-implementation-log.md # What was built and why
│ ├── 04-test-plan.md # Test strategy
│ ├── 04-test-results.md # Pass/fail results by test type
│ ├── 04b-threat-model.md # STRIDE threat analysis
│ ├── 04b-security-review.md # OWASP top 10 findings
│ ├── 05-deployment.md # Deployment guide
│ ├── 06-user-guide.md # End-user documentation
│ ├── 06-api-docs.md # API reference
│ └── 07-tech-debt.md # Unresolved issues (only if any remain)
├── src/ # Application source code
├── tests/ # Comprehensive test suite
│ ├── unit/ # Unit tests
│ ├── integration/ # Integration tests
│ ├── e2e/ # Playwright e2e tests
│ └── frontend/ # Frontend component tests
├── infra/ # Deployment config
│ ├── Dockerfile # Container definition
│ ├── docker-compose.yml # Local orchestration
│ └── .github/workflows/ # CI/CD pipeline
├── README.md # Project documentation
└── CLAUDE.md # Guidance for future Claude Code sessions
┌──────────────────────────────────────────────────────────────┐
│ THE FABRIC TEAM │
├──────────────┬──────────────┬──────────────┬─────────────────┤
│ fabric-pm │ fabric-ba │ fabric- │ fabric-ux │
│ Project │ Business │ architect │ UX/UI │
│ Manager │ Analyst │ Solution │ Engineer │
│ │ │ Architect │ │
├──────────────┼──────────────┼──────────────┼─────────────────┤
│ fabric- │ fabric-qa │ fabric- │ fabric-devops │
│ dev-lead │ QA │ security │ DevOps │
│ Lead │ Engineer │ Security │ Engineer │
│ Developer │ │ Engineer │ │
├──────────────┴──────────────┴──────────────┴─────────────────┤
│ fabric-docs │
│ Technical Writer │
└──────────────────────────────────────────────────────────────┘
- Fully autonomous — no human intervention after the initial prompt
- Full traceability — every requirement has an ID, traced through design, code, tests, and issues
- Parallel execution — architecture + UX run in parallel; QA + security run in parallel
- Quality gates — critical/high issues trigger automatic fix-and-retest loops (up to 3 rounds)
- E2E testing — Playwright tests for user flows, plus accessibility checks
- Security-first — STRIDE threat model + OWASP top 10 code review on every project
- Production-ready output — Dockerfile, CI/CD pipelines, deployment docs included
Fabric-specific code is MIT. Bundled third-party skills retain their own license files; see THIRD_PARTY_NOTICES.md.