Agent Claude Code
Software Architect — system design, design patterns, API design, database schema, and module boundaries. Invoked for structural decisions, refactoring, and new feature design.
Agent Claude Code
Code Reviewer — comprehensive code review for quality, readability, maintainability, and best practices. Invoked before any commit or PR.
Agent Claude Code
Chief Technology Officer — strategic technical decisions, architecture oversight, technology selection, and team coordination. Invoked for major architectural decisions, technology evaluations, and project-wide strategy.
Agent Claude Code
Data Scientist — data pipeline design, ML model architecture, statistical analysis, data validation, and performance metrics. Invoked for data-intensive features.
Agent Claude Code
Dependency Guardian — monitors and manages project dependencies. Checks for outdated packages, security vulnerabilities, unused dependencies, and license compliance. Invoked by /nightly-scan and /ship.
Agent Claude Code
DevOps Engineer — CI/CD pipelines, Docker, Kubernetes, infrastructure as code, monitoring, and deployment automation. Invoked for infrastructure and deployment tasks.
Agent Claude Code
Documentation Writer — API docs, README, architecture diagrams, onboarding guides, and inline documentation. Invoked after features are complete.
Agent Claude Code
Evolution Engine — the meta-agent that orchestrates self-improvement cycles. Combines scout findings + self-improver analysis + user feedback to evolve the entire system. This is the agent that makes all other agents better. Run monthly or when major updates are discovered.
Agent Claude Code
Meta Agent — creates new specialized subagents on demand. The agent that builds agents. Invoked when a new specialized capability is needed.
Agent Claude Code
Model Router — recommends which agent/model tier to use for a given task. Analyzes task complexity and suggests optimal agent. Invoked via /model-route.
Agent Claude Code
Multi-AI Orchestrator — routes tasks to the optimal AI model and coordinates cross-model workflows. Handles model selection, fallback chains, and result synthesis. Invoked when tasks benefit from multiple model perspectives or specific model strengths.
Agent Claude Code
Performance Engineer — profiling, optimization, caching strategy, database query tuning, and load testing. Invoked when performance is a concern or during optimization phases.
Agent Claude Code
QA Engineer — testing strategy, test writing, edge case discovery, regression testing, and quality gates. Invoked after any code change to validate quality.
Agent Claude Code
Technology Scout — searches GitHub, npm, and the web for new Claude Code features, MCP servers, agent patterns, and best practices. Keeps the setup current with the latest innovations. Run weekly or on-demand.
Agent Claude Code
Security Auditor — vulnerability scanning, OWASP compliance, auth/authz review, dependency auditing, secrets detection, OSINT reconnaissance, and security testing. Invoked on security-sensitive code changes and authorized penetration testing.
Agent Claude Code
Self-Improvement Agent — analyzes the entire Claude Code setup (agents, commands, hooks, CLAUDE.md) and finds ways to improve it. Scans for inefficiencies, missing patterns, outdated practices, and gaps. Then applies improvements. Run weekly or after every major project milestone.
Agent Claude Code
Swarm Leader — coordinates multi-agent teams, manages task boards, resolves conflicts, and ensures parallel work streams integrate cleanly. Invoked when running /swarm or when agent teams need coordination.
Command Claude Code
Execute a complete task from start to finish with ZERO human intervention. Claude expands the requirements, plans, builds, tests, and validates — autonomously.
Command Claude Code
For the given task, create 3 DIFFERENT implementations and pick the best one.
Command Claude Code
A structured, phase-based development workflow from product vision to deployed feature. Like a symphony conductor — each instrument plays at the right time.
Command Claude Code
Run a complete self-evolution cycle to improve the Claude Code setup.
Command Claude Code
The user is telling you about a mistake you just made. This is a learning opportunity.
Command Claude Code
Analyze the current task and recommend which agent (and therefore which model tier) to use. Model is set PER AGENT in YAML frontmatter — it cannot change mid-session.
Command Claude Code
Run a comprehensive project health check and generate a quality report.