The AI-Observable, Event-Native Modular Commerce Engine. Built for autonomous agents and modern developers.

Welcome to the home of Hyperrr, a next-generation distributed commerce engine designed from the ground up to be AI-observable, event-driven, and strictly modular.

Unlike legacy commerce systems, Hyperrr treats all business operations as deterministic, replayable DAG (Directed Acyclic Graph) workflows connected by a unified event fabric. It exposes dynamic capabilities to both human developers via GraphQL and autonomous AI agents via the Model Context Protocol (MCP).

               ┌─────────────────────────────────┐
               │    AI Agents & Web Clients      │
               └────────────────┬────────────────┘
                                │ (GraphQL / MCP)
                                ▼
               ┌─────────────────────────────────┐
               │         hyperrr Core            │
               │  (Workflow Engine + Event Bus)  │
               └────────┬────────┬────────┬──────┘
                        │        │        │
           ┌────────────┘        │        └────────────┐
           ▼                     ▼                     ▼
 ┌───────────────────┐ ┌───────────────────┐ ┌───────────────────┐
 │   commerce mod    │ │     auth mod      │ │   custom mod      │
 │  (Order/Cart/...) │ │ (APIKey/JWT/...)  │ │  (Your code)     │
 └─────────┬─────────┘ └─────────┬─────────┘ └─────────┬─────────┘
           │                     │                     │
           └─────────────────────┼─────────────────────┘
                                 │ (Decoupled Imports)
                                 ▼
                       ┌───────────────────┐
                       │        mdk        │
                       │ (Development Kit) │
                       └───────────────────┘

• 🤖 AI Agent Native: Integrated support for Model Context Protocol (MCP) allowing tools, workflows, and dashboard UIs to be rendered dynamically inside AI agent interfaces.
• ⚡ Event-Driven DAGs: Complex operations (like order fulfillment sagas) are modeled as declarative workflows with built-in rollback compensations.
• 🧩 Strict Compiler Decoupling: Functional modules compile and run independently of the core engine, utilizing go.work workspaces locally and standard Go module pseudo-versions in isolated CI pipelines.
• 📊 Visual Observability: Projector-based lineage tracking lets you trace every step of a workflow run in real-time.

• 🗃️ Out-of-Tree Database & Event-Bus Adapters: Decoupled the PostgreSQL GORM driver and NATS event bus/locker/workflow state store out of the core kernel into separate pluggable top-level modules (database and event-bus). Keeps core light and zero-dependency (defaulting to SQLite and in-memory event-bus).
• 🚀 Strategy 3 Workspace Dependency Resolution: Cleaned all go.mod files of local replace directives. Dependencies are now resolved sequentially using Git pseudo-versions and standard Go workspaces (go.work), optimizing package modularity and developer experience.
• 📁 Pluggable File Storage: Replaced the heavy cloud storage dependency in the core with a lightweight local disk and memory-based CloudProvider in the core engine. Out-of-tree cloud storage providers (S3, Azure Blob, and GCS) are now packaged under the separate file-storage repository.
• 🔑 Clean Interface-Based Actor Model: Decoupled the mdk.Actor definition into a pure interface, extracting the concrete database representation to auth.Actor inside the auth module.
• 🛠️ Dynamic Currency Support: Added dynamic currency configurations to hyperrr.yml and parameterized the MCP server to support formatting multiple currencies (e.g., USD, EUR, GBP, JPY, INR) dynamically based on catalog metadata and settings.
• 🔄 Parallel Workspace Testing: Restructured the workspace's CI/CD workflows across all repositories, resolving local replacement directives and running standalone generator bootstraps (build_ci.go) for parallel test execution.

Hyperrr uses Go Workspace (go.work) to link modules together locally for seamless editing.

1. Clone the entire stack:

   git clone https://github.com/GoHyperrr/hyperrr.git
   git clone https://github.com/GoHyperrr/mdk.git
   git clone https://github.com/GoHyperrr/commerce.git
   git clone https://github.com/GoHyperrr/auth.git
   git clone https://github.com/GoHyperrr/file-storage.git
   git clone https://github.com/GoHyperrr/database.git
   git clone https://github.com/GoHyperrr/event-bus.git

2. Setup the Go Workspace: Create a go.work file in your parent directory containing:

   go 1.25.5
   
   use (
       ./auth
       ./commerce
       ./database
       ./event-bus
       ./file-storage
       ./hyperrr
       ./mdk
   )

3. Run the server:

   cd hyperrr
   go run ./cmd/hyperrr server

We welcome contributions of all kinds! Whether you are writing new commerce adapters, enhancing the workflow orchestration engine, or building new AI tools, check out the Developer Guide to get started.

Let's build the future of agentic commerce together. ⚡