# Maestro

> Maestro is an open-source AI workflow skill with 25 commands, persistent memory, audit trail, and curated anti-patterns for building reliable, production-grade AI agent workflows. It provides workflow fluency — teaching AI agents how to behave across complex, multi-step tasks.

## Quick Start

- [Install Maestro](https://maestroskills.dev): `npx skills add sharpdeveye/maestro`
- [GitHub Repository](https://github.com/sharpdeveye/maestro): Source code, issues, and documentation
- [npm Package (MCP Server)](https://www.npmjs.com/package/maestro-workflow-mcp): MCP server for integration with any MCP-compatible client

## What Maestro Does

Maestro installs a meta-workflow layer that shapes how AI agents handle the entire lifecycle of any agentic task. It covers:

- **25 Workflow Commands** — planning, execution, checkpointing, task transitions, session memory, and handoffs
- **7 Reference Domains** — prompt engineering, context management, tool orchestration, agent architecture, feedback loops, knowledge systems, guardrails & safety
- **Persistent Memory Layer** — decisions, audit trail, and session history survive across sessions
- **Cost Estimation** — approximate cost tracking for Claude, GPT-4, Gemini, o3, o4-mini
- **Curated Anti-Patterns** — documented failure modes agents should avoid (context dumping, retry & pray, agent overkill, tool sprawl, ship & pray, no cost controls)
- **MCP Server** — dual transport (stdio + HTTP) with 10 tools for integration with 10+ providers

## Commands

### Analysis (3)
- `/diagnose` — Systematic workflow quality audit with scored dimensions
- `/evaluate` — Holistic review of workflow interaction quality
- `/reflect` — Analyze command history — which skills work, which fail, where to improve

### Fix & Improve (5)
- `/refine` — Final quality pass on prompts, tool descriptions, error messages
- `/streamline` — Remove unnecessary complexity, flatten pipelines
- `/calibrate` — Align workflow components to project conventions
- `/fortify` — Add error handling, retries, fallbacks, circuit breakers
- `/zero-defect` — Model-agnostic precision protocol with 8 execution rules

### Enhance (9)
- `/amplify` — Boost capabilities with better tools and context sources
- `/compose` — Design multi-agent orchestration and delegation patterns
- `/enrich` — Add knowledge sources and retrieval pipelines (RAG)
- `/accelerate` — Optimize for speed, reduce latency, minimize token usage
- `/chain` — Build effective tool chains and multi-step processes
- `/guard` — Add safety constraints, input validation, prompt injection defense
- `/iterate` — Set up feedback loops, evaluation cycles, self-correction
- `/temper` — Reduce over-engineering in multi-agent systems
- `/turbocharge` — Parallel orchestration, streaming, self-healing systems

### Utility (8)
- `/extract-pattern` — Extract reusable patterns from working workflows
- `/adapt-workflow` — Adapt workflows for different providers or environments
- `/onboard-agent` — Set up new agent configurations and bootstrap infrastructure
- `/specialize` — Make workflows domain-specific (legal, medical, financial, code)
- `/teach-maestro` — One-time context gathering, saves project context to .maestro.md
- `/capture` — Save a session summary — persist decisions, context, and what happened
- `/recap` — Quick summary of the last session — commands run, files changed, next steps

## MCP Server Integration

- **Local (stdio):** `npx -y maestro-workflow-mcp` — add to your MCP client config
- **Remote (HTTP):** `npx maestro-workflow-mcp --http --port 3001` — for shared teams, CI/CD, cloud environments
- **Exposes:** 25 prompts, 10 tools, 8 resources

## Compatible With

Claude Code, Cursor, VS Code (via Copilot or Antigravity), Gemini CLI, Antigravity, Kiro, Trae, OpenCode, Pi, and any MCP-compatible client

## Optional

- [Full Content](/llms-full.txt): Complete documentation for deep context
- [Sitemap](/sitemap.xml): XML sitemap for crawlers