rulem: AI Assistant Instruction Manager

A compact CLI for managing instruction files and MCP tooling across multiple repositories. Built with Go and Bubble Tea, rulem keeps each repository’s instructions, settings, and MCP tools in sync while sharing credentials, storage, and background scans.

Saving a rule file to the central repository — file picker with live markdown preview, then rename and save.

Quick orientation

• Multi-repo aware: Each repository gets its own instructions and settings inside the TUI, but all share the same credentials and MCP registry.
• Primary workflows: Launch rulem for the TUI, rulem mcp for the MCP server, and use the menu actions to save/import rules, refresh GitHub repos, or edit repository metadata.
• Safety guards: Git-backed flows run dirty-state checks before mutating branches, clone paths, or deleting a repo.

Quick start

1. Install (see below), then run rulem to launch the interactive UI.
2. Add or select a repository, use the main menu to save/import rules, or open the settings menu to edit metadata.
3. Run rulem mcp to expose your rule files as MCP tools for connected IDEs/assistants.
4. When developing, go test ./... and go run ./cmd/rulem are the canonical verification/build loops (see hacking.md for more).

Installation

• Homebrew (macOS): brew tap muhammadbassiony/rulem && brew install rulem
• Snap/Linux: sudo snap install rulem
• Binary: Download the latest release from GitHub Releases
• Source: git clone https://github.com/muhammadbassiony/rulem && go build -o rulem ./cmd/rulem

MCP integration

• Start the MCP server with rulem mcp (add --debug for verbose logging).
• Rule files with YAML frontmatter are auto-registered as MCP tools; each repo contributes tools that share the stored PAT/token.
• Use MCP inspectors (e.g., mcp-inspector) to confirm tool registration and invocation flows.