Hoofy is an MCP server written in Go that provides persistent memory, a structured spec-driven workflow, and an adaptive change management pipeline to help AI coding assistants build software more reliably. It exposes a suite of tools across four systems: Memory (persistent context with SQLite and FTS5), Change Pipeline (12 flow variants for evolving changes), Project Pipeline (greenfield spec-to-design workflow with a Clarity Gate and business rules extraction), and Bootstrap (reverse engineers existing codebases into SDD artifacts). You can interact with Hoofy via the MCP-compatible AI tools and agent prompts, leveraging the memory stores, governance gates, and on-demand prompts to drive consistent, spec-driven development. When running via Docker, Hoofy runs as a single binary inside the container, embedding SQLite at compile time for zero external dependencies. Use the provided memory and pipeline tools to store decisions, track changes, and validate specifications against your codebase before writing new code.

Prerequisites:

• Docker installed and running on your machine
• Basic familiarity with MCP-enabled AI tooling

1. Pull and run Hoofy in Docker:

   docker pull hendryavila/hoofy:latest docker run -it --rm
   -v /path/to/hoofy/data:/app/data
   hendryavila/hoofy:latest

   Notes:

   • The container stores memory and artifacts under /app/data by default. Mount a host directory to persist data.
   • The image exposes the Hoofy MCP server binary inside the container; no external dependencies are required.

2. Verify the server is accepting MCP connections:

   • Connect your MCP-enabled AI tooling to the Hoofy server endpoint (as configured in your environment). The container will listen on the default MCP port (adjust via environment variables if needed).

3. Alternative: Build from source (if you prefer to run the binary directly):

   go1.25+ install github.com/HendryAvila/Hoofy@latest

   Or clone the repo and build locally

   git clone https://github.com/HendryAvila/Hoofy.git cd Hoofy go build -o hoofy ./hoofy

Prerequisites for building locally: Go 1.25+ and a Go toolchain; ensure you have SQLite support compiled in.

Tips and common issues:

• Persistence: Hoofy uses an embedded SQLite database. Mount a host directory to /app/data to preserve memory across container restarts.
• Environment variables: You can tune persistence paths and port settings via environment variables exposed by the Docker image. Example: HOOFY_PERSISTENCE=/app/data/hoofy.sqlite
• On-demand prompts: Hoofy loads specific prompts for different stages (e.g., sdd-start, sdd-status, sdd-stage-guide) to minimize token usage while providing robust guidance.
• Compatibility: Works with Claude Code, Cursor, VS Code Copilot, Gemini CLI, OpenCode, and any MCP-compatible AI tool.
• Debugging: If the MCP endpoint cannot be reached, check Docker network settings, container logs, and ensure the host can reach the container port.
• Upgrades: When pulling a newer image, ensure data is persisted and consider taking a backup of the SQLite file before upgrading.