Summary

MulmoClaude provides a local, open-source platform for nurturing a deeply personalized AI assistant, ensuring user data sovereignty. It empowers users to build custom applications and integrate with communication channels, keeping all accumulated knowledge and data private. This project targets privacy-conscious users and developers interested in AI-native application architectures, offering a self-hosted, extensible AI companion.

How It Works

This AI-native application platform features Claude as a universal controller orchestrating plugins. It emphasizes a local-first, user-controlled approach, storing memories, data, and custom applications as plain files. Claude composes across plugins, and chat interactions dynamically summon GUIs like markdown, interactive charts, forms, or wikis, enabling no-code application development via natural language.

Quick Start & Requirements

• Primary Install/Run: npx mulmoclaude@latest
• Prerequisites: Node.js 20+, authenticated Claude Code CLI, ffmpeg.
• Optional: Docker Desktop (recommended for sandboxing), whisper.cpp (macOS only for local voice input).
• Docs: Developer Guide, Bridge Guide.

Highlighted Details

• Local Data Sovereignty: All user data, memories, and applications reside locally, ensuring privacy.
• AI-Native Platform: Claude orchestrates plugins for dynamic application building.
• No-Code App Development: Users describe desired applications in natural language for Claude to generate.
• Extensive Messaging Integration: Supports numerous chat platforms (Telegram, Slack, Discord, WhatsApp, Matrix, etc.) via dedicated bridges.
• Personal Wiki: Claude builds and maintains a markdown-based wiki for long-term memory.
• Dynamic Canvas: Chat results render as markdown, ECharts visualizations, spreadsheets, wikis, and more.
• Skill Execution: Integrates with Claude Code skills for running custom logic.
• Remote Access: Enables access from other devices via Google account authentication over Firebase.

Maintenance & Community

The README does not detail specific contributors, sponsorships, or community channels like Discord or Slack.

Licensing & Compatibility

• License: MIT.
• Compatibility: Permissive MIT license allows for commercial use and integration into closed-source projects.

Limitations & Caveats

• API Dependencies: Requires Gemini API key for image generation and X API token/credits for X integration.
• Security: Running without Docker exposes the user's filesystem to Claude Code's Bash access. The Docker sandbox mitigates this but may complicate user-installed skills due to symlink issues or CLI version mismatches.
• Platform-Specific Features: Local voice input is macOS-only and requires manual whisper.cpp compilation. PPTX attachment conversion relies on LibreOffice within the Docker sandbox.