Maestro App Factory
Multi-agent orchestration tool that uses AI to generate production-ready applications by organizing LLMs to operate like high-performing human development teams.
Philosophy
"The goal is production-ready apps, not just code snippets." Organizes agents into distinct roles mirroring professional software teams.
Core Agent Roles
- PM (Product Manager): Conducts interactive requirements interviews, adapts questions based on user expertise, reads existing codebases for context, generates requirements specifications.
- Architect: Transforms requirements into technical specs, breaks specs into stories, enforces engineering principles (DRY, YAGNI), reviews code, merges PRs -- but does not write code directly.
- Coders: Pull stories from queues, develop plans, implement code, run tests, submit PRs for architect review. They terminate and restart between stories.
Workflow
- PM conducts interview and generates spec
- Architect reviews and approves spec
- Architect breaks spec into stories and dispatches
- Coders plan, implement, and test
- Architect reviews code and merges PRs
- Coders terminate; new ones spawn for new work
Operating Modes
- Standard: Main workflow with GitHub integration
- Airplane Mode: Fully offline using local Gitea and Ollama
- Claude Code Mode: Uses Claude Code as subprocess
- Hotfix Mode: Express path for urgent production fixes
- Maintenance Mode: Automated technical debt management
Notable Features
- Knowledge graph capturing architectural patterns in
.maestro/knowledge.dot - Metrics dashboard tracking specs, stories, token usage, costs, test results
- State persistence via SQLite
- Mix-and-match LLM providers by agent type
Single binary (~15 MB), Go (95%), MIT license, Docker required.