Blueprinta is a high-fidelity, multi-agent orchestration platform designed to automate the end-to-end software development lifecycle. By coordinating specialized AI agents—Product Managers, Architects, Engineers, Security Experts, and DevOps—Blueprinta generates synchronized, production-ready system designs and implementation plans from simple natural language requests.
⚡ Quick Start • 📚 Documentation • 🤝 Contributing • 💬 Community • ⭐ Star us on GitHub
Building software today requires coordinating multiple specialized roles, each with their own expertise and tools. Blueprinta automates this entire process by:
- Eliminating manual coordination between PMs, Architects, Engineers, Security, DevOps, and QA
- Ensuring consistency across artifacts via tool-based refinement (Edit Artifacts API)
- Reducing development time from weeks to hours
- Providing production-ready outputs validated against industry standards
- Tool-based Refinement: Edit artifact JSON via predefined ops (
set_at_path,add_array_item, etc.) without re-running agents. UsePOST /api/diagrams/artifacts/editwith aneditsarray for deterministic, performant updates. - Context-Aware Agents: Each agent operates with deep awareness of upstream dependencies, ensuring architectural decisions are consistent across the stack.
- LLM-Native Caching: Leverages advanced Gemini context caching to minimize latency and token consumption during generation.
- Structured Validation: All agent outputs are validated against strict Zod schemas and JSON structures, ensuring reliable, machine-readable artifacts.
- Documentation Exports: Export documentation as Markdown, PDF, HTML, or PPTX via the export endpoint.
- Project Estimates: UI and exports include time, cost, and complexity estimates derived from artifacts.
- Draft Persistence: Create prompts, settings, and uploads persist across reloads on the create page.
- Agent-to-Agent (A2A) Communication: Specialized protocol for inter-agent delegation and task management.
- Generation Jobs + SSE: Long-running generation is queued, with progress streamed via Server-Sent Events.
- Guest Support: Limited diagram generation for non-authenticated users with session tracking.
- Gemini-Powered: Built on Google Gemini with context caching for fast, high-quality structured output.
Blueprinta utilizes a sequential pipeline:
- Product Manager: Defines user stories and acceptance criteria.
- Architect: Generates API contracts, database schemas, and ADRs.
- DevOps Infrastructure: Designs CI/CD pipelines and cloud infrastructure (Terraform/K8s).
- Security Architecture: Performs threat modeling and defines security controls.
- UI Designer: Generates design tokens and component hierarchies.
- Engineer Implementation: Drafts file structures and technical implementation plans.
- QA Verification: Creates comprehensive test strategies and test cases.
Get Blueprinta up and running in under 5 minutes:
# Clone the repository
git clone https://github.com/josephsenior/Metasop.git
cd Metasop
# Install dependencies
pnpm install
# Set up environment variables
cp .env.example .env
# Edit .env and add your Gemini API key (GOOGLE_AI_API_KEY)
# Create local database (SQLite)
pnpm db:generate
pnpm db:push
# Run the development server
pnpm dev
# Open http://localhost:3000- Node.js 18+ (Download)
- pnpm (recommended) or npm/yarn
- Gemini API Key (Get one here)
- Git for version control
Create a .env file in the root directory (see .env.example for a template):
# Required: Gemini API Key (get one at https://aistudio.google.com/apikey)
GOOGLE_AI_API_KEY=your_gemini_api_key_here
# Required for local storage: SQLite
DATABASE_URL="file:./prisma/local.db"
# Optional: LLM Provider (default: gemini)
METASOP_LLM_PROVIDER=gemini
# Optional: use mock provider for offline/dev tests
# METASOP_LLM_PROVIDER=mock
# Optional: Model (default: gemini-3-flash-preview)
METASOP_LLM_MODEL=gemini-3-flash-preview
# Optional: Agent timeout (ms) and retries
# METASOP_AGENT_TIMEOUT=180000
# METASOP_AGENT_RETRIES=0# Run integration tests (orchestration flow)
npx tsx tests/integration/verify_full_pipeline.ts
npx tsx tests/integration/test_cascading_refinement.ts
# With custom model
METASOP_LLM_MODEL=gemini-3-pro-preview npx tsx tests/integration/verify_full_pipeline.ts# Start development server
pnpm dev
# Type checking
pnpm type-check
# Linting
pnpm lint
pnpm lint:fix
# Build for production
pnpm build
pnpm startOn Windows, pnpm build may fail with an EPERM symlink error when Next.js generates .next/standalone. See docs/TROUBLESHOOTING.md.
The project uses Vitest for unit tests. The suite covers the Blueprinta pipeline: orchestrator, agents, services, adapters, and utilities.
pnpm test # Run unit tests once
pnpm test:watch # Run tests in watch mode
pnpm test:coverage # Run tests with coverage report (lib/, components/, app/api/)
pnpm test:ui # Open Vitest UIIntegration tests under tests/integration/ are excluded from the default run and can be executed separately. Coverage reports are written to ./coverage (HTML, LCOV, text).
If pnpm test fails with spawn EPERM in Cursor's terminal (common on Windows), see Tests: spawn EPERM.
| Guide | Description |
|---|---|
| Documentation Hub | Start here — full index and learning path |
| Setup Guide | Install, configure, and run locally |
| Architecture | System design and agent pipeline |
| API Reference | REST endpoints and examples |
| LLM Configuration | Gemini setup, model selection, optimization |
| Deployment | Deploy to Vercel, Docker, VPS |
| Testing | Unit and integration testing |
| Troubleshooting | Common issues and fixes |
| Contributing | How to contribute |
Blueprinta utilizes a sequential pipeline with feedback loops:
graph LR
A[User Request] --> B[Product Manager]
B --> C[Architect]
C --> D[DevOps]
D --> E[Security]
E --> F[UI Designer]
F --> G[Engineer]
G --> H[QA Verification]
- Product Manager: Defines user stories and acceptance criteria.
- Architect: Generates API contracts, database schemas, and ADRs.
- DevOps Infrastructure: Designs CI/CD pipelines and cloud infrastructure (Terraform/K8s).
- Security Architecture: Performs threat modeling and defines security controls.
- UI Designer: Generates design tokens and component hierarchies.
- Engineer Implementation: Drafts file structures and technical implementation plans.
- QA Verification: Creates comprehensive test strategies and test cases.
- Orchestrator: Runs agents in order and passes artifacts as context; no graph in DB.
- Execution Service: Handles timeouts, retries, and error handling.
- Refinement: Tool-based only via
POST /api/diagrams/artifacts/edit(no agent re-runs).
We love contributions! Whether you're fixing a bug, adding a feature, or improving documentation, we want your help.
- Fork the repository and create your branch (
git checkout -b feature/amazing-feature) - Make your changes and ensure tests pass (
pnpm test) - Commit your changes (
git commit -m 'Add some amazing feature') - Push to the branch (
git push origin feature/amazing-feature) - Open a Pull Request
For detailed guidelines, see CONTRIBUTING.md.
# Fork and clone the repository
git clone https://github.com/YOUR_USERNAME/Metasop.git
cd Metasop
# Install dependencies
pnpm install
# Create a feature branch
git checkout -b feature/my-feature
# Make your changes and test
pnpm test
pnpm type-check
pnpm lint
# Commit and push
git add .
git commit -m "feat: add my feature"
git push origin feature/my-feature- GitHub Discussions: Join the conversation
- Discord Server: Join our Discord (coming soon)
- Twitter/X: Follow @MetaSOP_AI for updates
- Blog: Read our latest posts (coming soon)
- 📖 Check the Documentation
- 🐛 Report bugs on GitHub Issues
- 💡 Ask questions on GitHub Discussions
- 💬 Join our Discord community
- Multi-agent orchestration system
- Tool-based refinement (Edit Artifacts API)
- Knowledge graph for dependency tracking
- Agent-to-agent communication protocol
- Web interface with Next.js
- Additional LLM providers (OpenAI, Anthropic, etc.)
- SSE progress streaming for generation
- Custom agent templates marketplace
- Advanced analytics and insights
- Mobile app
- Enterprise features (SSO, audit logs, etc.)
See ROADMAP.md for detailed plans.
Support Blueprinta's development by becoming a sponsor! Your support helps us:
- Maintain and improve the core platform
- Add new features and integrations
- Provide better documentation and support
- Keep the project free and open source
Distributed under the MIT License. See LICENSE for more information.
- Built with Next.js
- Powered by Google Gemini
- UI components from Radix UI
- Styled with Tailwind CSS
- Testing with Vitest
Built with ❤️ by the Blueprinta community
⭐ Star us on GitHub • 🐛 Report a Bug • 💡 Request a Feature
Automating the future of engineering