Skip to main content

Quick Start

Get running in 30 seconds.

Install

npm install -g specweave

Starting a New Project (Greenfield)

The simplest way to start — just describe what you want to build:

mkdir my-app && cd my-app
specweave init .

Then in Claude Code, simply say:

"Build a calculator app with React"

SpecWeave guides you through features, tech stack, and approach — then creates your first increment automatically.

Perfect for prototypes, learning, and weekend MVPs.


Adding Features to Existing Projects

For existing codebases, use explicit commands:

cd your-project
specweave init .

Then in Claude Code:

/sw:increment "Add dark mode toggle"

SpecWeave creates:

.specweave/increments/0001-dark-mode/
├── spec.md <- WHAT: User stories + acceptance criteria
├── plan.md <- HOW: Architecture + tech decisions
└── tasks.md <- DO: Tasks with embedded tests

Execution Options

Ship While You Sleep (Auto Mode)

For autonomous execution that can run for hours:

/sw:auto

Auto mode executes tasks, runs tests, fixes failures, and syncs to GitHub/JIRA — completely hands-off. Check progress with /sw:auto-status or stop with /sw:cancel-auto.

Step-by-Step Control

For manual control:

/sw:do           # Execute one task
/sw:done 0001 # Quality-validated completion

Pro tip: Use /sw:next to flow through the entire cycle. One command auto-closes completed work and suggests what's next — review specs/tasks when needed, otherwise just keep clicking "next".

Your specs, architecture, and tests are now permanent documentation.

💡 VSCode Users: Use Compact Mode!

Claude Code now supports compact mode — keep the AI assistant inside your editor window instead of a separate terminal.

Just type: compact

This lets you work continuously for hours in the same VSCode window without context switching. Combined with /sw:auto, you can ship entire features while focusing on other work.

Bonus: Stop hooks now work with subagents, so SpecWeave's quality gates validate every level of autonomous execution automatically.


When to Use Increments

Not every change needs an increment. The rule of thumb:

Change TypeUse Increment?Why
Typo fix, version bumpNoZero learning, purely mechanical
Bug fix that taught you somethingYesCapture the knowledge for future devs
Any user-facing changeYesTrack delivery, enable DORA metrics
Architecture decisionYesNeeds ADR for future understanding

The principle: If you'd explain this change to a colleague, document it in an increment.

Increments capture knowledge. Ad-hoc work is ephemeral. Quick check before doing ad-hoc work: "Will I remember why I made this change in 6 months?" If no → create an increment.


What You Get

After specweave init .:

ComponentCountPurpose
Skills136Auto-activating capabilities (planning, TDD, brownfield, sync)
Agents68Specialized roles (PM, Architect, DevOps, QA, Security, SRE)
Commands53Slash commands for workflow control
Hooks3+Event-driven automation (lifecycle, sync, status)
CLAUDE.md1Your project reference guide
Context Efficiency with MCP Tool Search

Claude Code 2.1.7+ includes MCP Tool Search (lazy loading) — SpecWeave's 24 plugins load on-demand instead of all at once.

Result: 85%+ context reduction (~100k → ~5-10k tokens at startup), enabling longer /sw:auto sessions and more complex increments.

No configuration needed — enabled by default when MCP tools exceed 10% of context.

Coming in v1.1: Native Lazy Plugin Loading — 99% token savings for non-SpecWeave work via router-based architecture.

Recommended MCP Servers

Two MCP servers that supercharge SpecWeave:

Context7 — Real-time docs for any library (no more hallucinated APIs):

claude mcp add context7 -- npx -y @anthropic-ai/context7-mcp

Playwright — Browser automation for E2E testing (Claude sees your app):

claude mcp add playwright -- npx -y @anthropic-ai/playwright-mcp

These add capabilities CLI tools can't match — current documentation and visual verification.


Essential Commands

CommandPurpose
/sw:increment "..."Create new feature with AI agents
/sw:auto🚀 Ship while you sleep - hours of autonomous work
/sw:doExecute one task at a time
/sw:done <id>Complete with quality gates
/sw:nextAuto-close + suggest next (one-click flow)
/sw:progressCheck status
/sw:auto-statusCheck autonomous session progress
/sw:cancel-autoStop autonomous session
/sw:sync-progressSync to GitHub/JIRA/ADO
/sw:sync-monitorDashboard: jobs, notifications
/sw:discrepanciesView code-to-spec drift

Example: Build an Event Management SaaS

# Install
npm install -g specweave

# Create project
mkdir eventmgmt && cd eventmgmt
specweave init .

# Open Claude Code and describe:
"Build an event management SaaS with Next.js 14, Prisma, NextAuth.js,
Stripe payments, deployed on Hetzner Cloud"

# SpecWeave autonomously creates:
# - PRD with market research
# - Architecture with C4 diagrams
# - Database schema (Prisma)
# - Auth system (NextAuth.js)
# - Payment integration (Stripe)
# - Infrastructure (Terraform for Hetzner)
# - Deployment pipeline ([GitHub Actions](/docs/glossary/terms/github-actions))
# - Tests ([Playwright](/docs/glossary/terms/playwright) [E2E](/docs/glossary/terms/e2e) + Jest)
# - Living documentation (auto-updates)

# Then say: "Implement the MVP"
# SpecWeave builds the entire application!

Joining an Existing Project (Brownfield)

If you're working with legacy code, SpecWeave can analyze your codebase for documentation gaps:

specweave init .
# During init, select "Run brownfield analysis"

After analysis completes:

/sw:discrepancies                    # View all documentation gaps
/sw:discrepancies --module auth # Filter by module
/sw:discrepancy-to-increment DISC-0001 DISC-0002 # Create increment

The background analysis runs while you continue working. Check progress with:

/sw:jobs

Sync Monitoring (NEW)

SpecWeave provides sync monitoring for external tools (GitHub/JIRA/ADO):

/sw:sync-progress    # Push updates to external tools
/sw:sync-monitor # See sync status dashboard
/sw:notifications # View/dismiss sync alerts

Code is the source of truth. When docs drift from reality, you get notified:

/sw:discrepancies    # See code-to-spec drift
/sw:discrepancies accept DISC-0001 # Update specs from code

Configuration (Optional)

Edit .specweave/config.yaml:

project:
name: "your-project"
type: "greenfield" # or "brownfield"

hooks:
enabled: true
post_task_completion:
enabled: true

testing:
e2e_playwright_mandatory_for_ui: true
min_coverage: 80

integrations:
github:
enabled: true
jira:
enabled: false

Customization (CLAUDE.md)

SpecWeave is a framework, not a locked product. Customize workflows through your project's CLAUDE.md:

# In your CLAUDE.md:

## Custom Sync Rules
When syncing to JIRA, always:
- Add custom field "Team: Backend" for backend increments
- Map "paused" status to "Blocked" instead of "On Hold"

## Custom Quality Gates
Before closing any increment:
- Run `npm run lint:strict` in addition to tests
- Verify changelog entry exists

## Custom Workflow
After completing a task:
- Post to #dev-updates Slack channel

What you can customize:

  • External sync — Add fields, transform statuses, integrate with internal tools
  • Quality gates — Add custom validation, linting, security scans
  • Lifecycle hooks — Trigger actions on increment events
  • Agent behavior — Override agent prompts for your domain
  • Naming conventions — Enforce team-specific formats

Requirements

Node.js Version Required

SpecWeave requires Node.js 20.12.0 or higher (we recommend Node.js 22 LTS).

Check your version: node --version

If you see an error like SyntaxError: Unexpected token 'with', your Node.js is too old. See how to upgrade.

  • Node.js 20.12.0+ (node --version) — upgrade instructions
  • npm 9+ (npm --version)
  • Claude Code (recommended) or any AI tool
  • Git (for version control)

Troubleshooting

Quick Recovery (Most Issues)

If commands, skills, or hooks stop working:

specweave update      # Full update: CLI + plugins + instructions

This fixes 98% of issues. It updates everything: CLI version, plugins, instruction files.

Auto Mode Issues

Session stuck or not completing?

/sw:auto-status   # Check what's happening
/sw:cancel-auto # Cancel if needed
/sw:auto # Resume with fresh session

Skills or Hooks Not Working?

First, try the full update:

specweave update

If that doesn't help, use the plugin-only refresh:

specweave refresh-marketplace

Why this exists: Claude Code's native marketplace auto-update doesn't:

  • Fix hook permissions (chmod +x) — hooks may silently fail
  • Clean up orphaned cache/skills directories
  • Update instruction files (CLAUDE.md, AGENTS.md)

Enable Marketplace Auto-Update (Optional)

You can enable Claude Code's native auto-update for the specweave marketplace:

/plugin → Marketplaces tab → Select specweave → Enable auto-update

This keeps marketplace.json and installed plugins updated automatically. However, you may still need specweave refresh-marketplace occasionally for hook permissions and instruction file updates.


Next Steps


Ready to build permanent knowledge instead of losing work to chat history?

npm install -g specweave && specweave init .