Commands Reference

The AI Context System v5.0.2 uses an agent-based architecture:

• 22 Slash Commands - Entry points for all functionality
• 12 Specialist Agents - Domain experts (security reviewer, performance reviewer, etc.)
• 7 JSON Schemas - Contract validation for all agents
• Session Hooks - Automatic context health checks

Commands are organized into categories below.

Setup Commands (Run Once)

/init-context

Creates fresh context structure for new projects. Best for projects starting from scratch.

/migrate-context

Migrates existing documentation to AI Context System structure. Preserves all existing content.

Daily Use Commands

/save

Quick save (2-3 min) - Updates STATUS.md with auto-generated Quick Reference. Run frequently during work.

/save-full

Comprehensive save (10-15 min) - Everything /save does + SESSIONS.md entry with mental models. Run before breaks/handoffs.

/review-context

Run at session start - Verifies documentation is current. Shows Quick Reference. Checks for updates.

Code Review Commands

Agent-Based Architecture (v5.0.0)

The code review system now uses specialist agents with self-declaring contracts. Each review domain is handled by a dedicated agent in .claude/agents/ that declares its own capabilities via JSON Schema-validated contracts.

/code-review

Master orchestrator - Select and run specialized audits via interactive menu or command-line arguments. Supports presets like --prelaunch, --backend, --frontend.

/code-review-security

OWASP Top 10 security audit. Authentication, injection, XSS, secrets, dependencies.

/code-review-performance

Core Web Vitals audit. LCP, INP, CLS, bundle size, caching.

/code-review-accessibility

WCAG 2.1 AA compliance audit. Keyboard navigation, screen readers, ARIA.

/code-review-seo

Technical SEO audit. Metadata, structured data (JSON-LD), crawlability.

/code-review-database

Database efficiency audit. N+1 detection, indexes, connection pooling.

/code-review-infrastructure

Serverless cost optimization. Cold starts, rendering strategy, caching.

/code-review-typescript

TypeScript type safety audit. Strict mode, any tracking, type coverage.

/code-review-testing

Test coverage and quality audit. Coverage metrics, test pyramid, CI integration.

/build-check

Pre-push build gate. Sequential: lint → typecheck → tests → build.

Collaboration Commands

/export-context

Package all context for handoffs. Combines documentation into single markdown file.

/validate-context

Check documentation health. Reports staleness, missing sections, and health score.

Maintenance Commands

/update-context-system

Updates commands and scripts from GitHub. Ensures latest features and fixes.

/update-templates

Compare context files with latest templates. Interactive updates with visual diffs.

/add-ai-header

Create AI tool-specific entry points (CLAUDE.md, cursor.md, etc.).

/organize-docs

Interactive documentation cleanup and organization.

/session-summary

Generate summary of current session for handoff.

Command Quick Reference

Category	Commands
Setup	/init-context, /migrate-context
Daily	/save, /save-full, /review-context
Code Review	/code-review (orchestrator) + 8 specialized domain audits
Collaboration	/export-context, /validate-context
Maintenance	/update-context-system, /update-templates, /add-ai-header, /organize-docs, /session-summary

Command Philosophy

All commands follow these principles:

1. Non-destructive - Never overwrite without confirmation
2. Transparent - Show exactly what will change
3. Auditable - Log all operations
4. Resumable - Safe to interrupt and restart
5. Helpful - Provide guidance and next steps

Architecture (v5.0.2)

The system uses an agent-based architecture:

.claude/
├── commands/    # 22 slash commands
├── agents/      # 12 specialist agents (code reviewers, etc.)
├── schemas/     # 7 JSON Schema contracts
└── hooks/       # Session automation

For Claude Code users: Use slash commands directly (e.g., /save)

For other AI tools: Reference agent files in .claude/agents/ to follow the same workflows