tribal-memory/tribal

What it does

Tribal is a self-hosted semantic memory server designed specifically for engineering teams to preserve and query tribal knowledge—the implicit decisions, reasoning, and hard-won insights embedded in code and team practices. It provides a persistent, queryable knowledge base that can be accessed by both team members and AI agents through the Model Context Protocol (MCP).

How to set up

Installation requires: (1) Cloning the repository from tribalmemory/tribal on GitHub, (2) Setting up a PostgreSQL database with pgvector extension enabled, (3) Configuring connection parameters, (4) Running the MCP server locally or on your infrastructure, (5) Connecting it to your Claude, ChatGPT, or other MCP-compatible AI tools via the server endpoint.

Key features

• Semantic search powered by pgvector for intelligent knowledge retrieval
• Self-hosted architecture for complete data control and privacy
• MCP (Model Context Protocol) integration for seamless AI agent access
• PostgreSQL backend for reliable, scalable storage
• Built with Rust for performance and reliability
• Team-focused design capturing collective engineering wisdom
• Persistent memory layer that improves over time with additions