Skip to content

Privacy-First Recommendations for the Fediverse

Corgi brings intelligent content discovery to Mastodon without compromising privacy. Built by the community, for the community.

- :material-clock-fast: **[Start in 5 Minutes](get-started/02_quick-integration.md)** --- Deploy Corgi with Docker and make your first API call - :material-book-open-variant: **[Read the Concepts](concepts/01_architecture-deep-dive.md)** --- Understand the architecture and design principles - :material-code-braces: **[API Reference](reference/api/index.md)** --- Complete API documentation with examples - :material-heart: **[Join Us](development/01_contributing.md)** --- Contribute to the future of ethical recommendations

The Corgi Ecosystem

graph LR
    subgraph Users ["Fediverse Users"]
        U1["New User"]
        U2["Active User"]
        U3["Power User"]
    end

    subgraph Clients ["Mastodon Clients"]
        ELK["ELK Client"]
        WEB["Web UI"]
        MOBILE["Mobile Apps"]
    end

    subgraph Corgi ["Corgi Platform"]
        API["REST API"]
        REC["Recommendation Engine"]
        PRIVACY["Privacy Layer"]
    end

    subgraph Backend ["Your Infrastructure"]
        MAST["Mastodon Instance"]
        DB[("Database")]
    end

    U1 --> ELK
    U2 --> WEB
    U3 --> MOBILE

    ELK --> API
    WEB --> API
    MOBILE --> API

    API --> REC
    API --> PRIVACY

    REC --> MAST
    PRIVACY --> DB

    classDef users fill:#e3f2fd,stroke:#1976d2,color:#000
    classDef clients fill:#f3e5f5,stroke:#7b1fa2,color:#000
    classDef corgi fill:#e8f5e8,stroke:#388e3c,color:#000
    classDef backend fill:#fff3e0,stroke:#f57c00,color:#000

    class U1,U2,U3 users
    class ELK,WEB,MOBILE clients
    class API,REC,PRIVACY corgi
    class MAST,DB backend

Why Corgi?

Privacy Without Compromise

Unlike traditional recommendation engines, Corgi never tracks users across the web. Your data stays on your instance, and recommendations are generated using privacy-preserving techniques like user aliasing and local processing.

Production-Ready Performance

With response times under 100ms and the ability to handle 2.5M+ daily requests, Corgi scales with your community. Advanced caching and asynchronous processing ensure smooth operation even during peak loads.

Developer-Friendly Integration

Whether you're building a new Mastodon client or enhancing an existing one, Corgi's comprehensive API and SDKs make integration straightforward. Full OpenAPI documentation and client libraries get you started quickly.

Community-Driven Development

Open source from day one, Corgi is built by the Fediverse community. Every algorithm is transparent, every decision is documented, and every contribution makes the platform better for everyone.

Multi-Algorithm Recommendations

Corgi combines multiple recommendation strategies to provide diverse, relevant content:

  • Semantic Understanding: Find similar content even with different wording
  • Social Graphs: Leverage community connections
  • Temporal Relevance: Balance fresh and evergreen content
  • Engagement Prediction: Learn from user interactions

Enterprise-Grade Features

  • A/B Testing Framework: Experiment with recommendation strategies
  • RAG-Powered Development: AI-assisted code understanding
  • Agent Orchestration: Coordinate complex AI workflows
  • Comprehensive Monitoring: Prometheus metrics and Grafana dashboards

Privacy-First Architecture

  • User Aliasing: One-way hashing protects identities
  • Local Processing: Sensitive data never leaves your instance
  • Data Minimization: Only collect what's necessary
  • Transparent Algorithms: No black boxes or dark patterns

Getting Started

Ready to enhance your Fediverse experience?

- :material-information: **[What is Corgi?](get-started/01_what-is-corgi.md)** --- Learn about our mission and approach - :material-docker: **[Quick Start Guide](get-started/02_quick-integration.md)** --- Deploy in minutes with Docker - :material-school: **[Browse All Guides](guides/01_elk-integration.md)** --- Explore integration options

Repository: GitHub • Issues • Discussions