Claude Code Basics
Claude Code is the AI coding assistant that SpecWeave is built on. Understanding its core concepts helps you get the most out of SpecWeave.
While Claude Code provides the deepest integration (native hooks, skills, agents), SpecWeave's core workflow also works with Cursor, Windsurf, GitHub Copilot, and any AI IDE via CLAUDE.md instructions.
What is Claude Code?
Claude Code is Anthropic's CLI-based AI coding assistant. Unlike chat-based tools (ChatGPT, Claude.ai), Claude Code:
- Runs inside your terminal, with direct access to your codebase
- Can read, write, and edit files directly
- Can run commands (tests, builds, git operations)
- Supports plugins that extend its capabilities
- Works with the latest Claude models (Opus 4.6, Sonnet 4.6, Haiku 4.5)
# Install Claude Code
npm install -g @anthropic-ai/claude-code
# Start in your project directory
cd your-project
claude
Key Concepts
Models
Claude Code supports multiple AI models with different strengths:
| Model | Best For | Speed | Cost |
|---|---|---|---|
| Opus 4.6 | Complex architecture, multi-file changes, autonomous work | Slower | Higher |
| Sonnet 4.6 | Balanced daily coding, reviews, refactoring | Medium | Medium |
| Haiku 4.5 | Quick lookups, simple edits, documentation | Fast | Low |
Recommendation: Use Opus 4.6 for SpecWeave workflows (especially sw:auto). Its reasoning depth is critical for multi-step planning and implementation.
Plugins
Plugins are extension packages that add capabilities to Claude Code. SpecWeave is delivered as a set of plugins.
Claude Code (base)
└── Plugins
└── specweave ← Unified plugin
├── Core workflow (increment, do, auto, done)
├── GitHub integration (Issues, PRs, milestones)
├── JIRA integration (epics, stories, sprints)
└── ... (all integrations bundled)
Plugins load on-demand based on what you're doing. Say "React frontend" and the frontend plugin activates. Say "Kubernetes deploy" and the infrastructure plugin loads. No manual configuration needed.
Skills
Skills are auto-activating expertise modules inside plugins. You can trigger them in two ways:
Natural language -- just describe what you need. SpecWeave detects your intent and activates the right skill:
You: "Design the authentication architecture"
↓
Claude Code detects: "architecture" keyword
↓
sw:architect skill activates → System design expertise loads
Slash commands -- for explicit control:
Other examples: sw:pm (product management), sw:security (security audit).
Most users never need slash commands. Describing what you want is enough -- SpecWeave routes to the right skill automatically.
SpecWeave has ~48 built-in skills covering PM, architecture, security, testing, frontend, backend, DevOps, and more.
For a deep dive into how skills, plugins, and marketplaces work, see the Fundamentals guide.
Agents
Agents are specialized subprocesses that handle specific types of work. Think of them as team members with expertise:
| Agent | What It Does |
|---|---|
| PM Agent | Writes user stories, defines acceptance criteria |
| Architect Agent | Designs systems, writes ADRs, picks technology |
| QA Lead Agent | Creates test strategies, defines quality gates |
| Security Agent | Reviews for OWASP vulnerabilities, threat models |
| Frontend Agent | React/Vue/Angular architecture and patterns |
| Backend Agent | API design, database optimization, .NET/Node.js patterns |
22+ agents work together. They activate based on context - you don't need to manage them manually.
Hooks
Hooks are automated actions that fire in response to events. SpecWeave uses hooks to enforce quality:
| Hook | When It Fires | What It Does |
|---|---|---|
task-ac-sync | After completing a task | Updates acceptance criteria in spec.md |
tdd-enforcement | Before marking task complete | Ensures tests were written first |
grill-gate | Before closing increment | Ensures code review was done |
living-specs | After task completion | Updates living documentation |
Hooks run automatically. You don't invoke them - they enforce quality standards behind the scenes.
Three Invocation Methods
| Method | How It Works | Example |
|---|---|---|
| Natural language | Describe what you want -- SpecWeave detects intent | "I want to add dark mode" |
| Slash commands | Explicit Claude Code commands with sw: prefix | sw:increment "dark mode" |
| Keywords | Same command names without prefix (other AI tools) | increment "dark mode" |
All three trigger the same underlying skills and produce identical results.
Common Activation Phrases
| What You Want | Natural Language | Slash Command |
|---|---|---|
| Plan a feature | "Let's build [feature]" | sw:increment |
| Start building | "Start implementing" | sw:do |
| Check progress | "What's the status?" | sw:progress |
| Go autonomous | "Ship while I sleep" | sw:auto |
| Finish up | "We're done" | sw:done |
| Brainstorm | "Let's brainstorm ideas" | sw:brainstorm |
How SpecWeave Uses Claude Code
SpecWeave layers a spec-driven workflow on top of Claude Code's capabilities:
Without SpecWeave: Claude Code is a powerful AI assistant that loses context when the chat ends.
With SpecWeave: Claude Code becomes an AI development team with permanent memory, quality gates, and external tool sync.
Why Claude Code Over Other Tools?
SpecWeave works with other AI tools, but Claude Code provides the deepest integration:
| Capability | Claude Code | Other AI Tools |
|---|---|---|
| Native plugins | Full plugin system | Via CLAUDE.md instructions |
| Hooks (automation) | Pre/post tool hooks | Not available |
| Skill auto-activation | Keyword-based loading | Manual prompting |
| Agent spawning | Native subagent support | Not available |
| Terminal access | Direct command execution | Varies by tool |
| File operations | Read, write, edit directly | Varies by tool |
Bottom line: Claude Code + SpecWeave = native integration. Other tools + SpecWeave = works, but without hooks and auto-activation.
Next Steps
- Install SpecWeave - Set up your project
- Claude Code Architecture - Deep dive into the plugin system
- Academy - 16-lesson learning path
- All 100+ Skills - Browse available expertise