The Model Context Protocol (MCP) is a standardized protocol that allows AI assistants like Claude to interact with external tools and data. BigRack implements this protocol to provide persistent, structured context to AI development tools.
Via MCP, Claude can query BigRack to get business context, plan tasks, and validate that code respects defined rules.
BigRack Installation
│
├── Repo: "E-commerce Platform"
│ │
│ ├── 📚 Business Context (shared)
│ │ ├── Business Rules
│ │ ├── Glossary Entries
│ │ ├── Architecture Patterns
│ │ └── Team Conventions
│ │
│ └── 📋 Projects
│ ├── Project: "Payment Integration"
│ │ ├── Task: "Setup payment gateway API keys"
│ │ ├── Task: "Implement payment processing endpoint"
│ │ ├── Task: "Add payment validation logic"
│ │ └── Task: "Write payment integration tests"
│ ├── Project: "Fix Cart Bug"
│ │ ├── Task: "Reproduce the cart bug"
│ │ ├── Task: "Identify root cause"
│ │ ├── Task: "Implement fix"
│ │ └── Task: "Add regression tests"
│ └── Project: "Add Tests"
│ ├── Task: "Setup test framework"
│ ├── Task: "Add unit tests for core modules"
│ └── Task: "Add integration tests"
│
└── Repo: "Marketing Website"
└── ...A Repo represents a business domain or codebase. It's the main container for your business context.
A Project represents a specific work unit(feature, bugfix, refactor, etc.).
A Ticket represents an atomic task within a project. It's the smallest unit of work that can be planned, tracked, and completed independently.
Business context is the structured knowledge that allows BigRack to understand your domain and validate that your code respects your rules.
Business rules that must be respected by the code.
Example: "Stock cannot be negative", "Discounts max 70%"
Definitions of terms specific to your domain.
Example: "SKU: Stock Keeping Unit", "Cart: Shopping basket"
Patterns and architectural decisions to follow.
Example: "Use Repository pattern", "NestJS for backend"
Code conventions and team standards.
Example: "Test coverage > 80%", "Use Prettier"
BigRack can decompose a complex feature into atomic tickets with their dependencies. This creates a dependency graph (DAG) that guides development.
Each ticket can have validation criteria, objectives, and be linked to a Git branch.
BigRack uses vector embeddings and semantic search to find relevant context for your queries. This enables a RAG (Retrieval-Augmented Generation) workflow.