Summary

Memoh is an always-on, containerized AI agent system designed for creating and managing multiple, isolated AI bots. It targets power users and engineers seeking a stable, secure, and feature-rich platform for personal assistants, household management, or collaborative bot teams. Memoh offers persistent memory and cross-platform interaction, addressing limitations of other agent frameworks.

How It Works

The system leverages Containerd to provide robust isolation for each AI bot, treating them like independent machines with dedicated compute and persistent storage. It employs a sophisticated memory architecture combining hybrid retrieval (dense vector search and BM25 keyword search) with LLM-driven fact extraction. Bots can execute commands, edit files, and interact seamlessly across various communication channels, remembering context per user and bot.

Quick Start & Requirements

• Installation: One-click install script (curl -fsSL https://raw.githubusercontent.com/memohai/Memoh/main/scripts/install.sh | sudo sh) or manual Docker Compose setup (git clone --depth 1 https://github.com/memohai/Memoh.git && cd Memoh && sudo docker compose up -d).
• Prerequisites: Docker.
• Access: Web UI available at http://localhost:8082 (default login: admin/admin123).
• Resources: See DEPLOYMENT.md for custom configuration and production setup.
• Links: [Telegram Group], [Documentation], [Cooperation].

Highlighted Details

• Multi-Bot & Multi-User: Manage numerous bots and recognize individual users across group chats, with role-based access control.
• Containerized Isolation: Each bot runs in its own containerd container, supporting snapshots for save/restore.
• Advanced Memory: Hybrid retrieval (vector + BM25) and LLM fact extraction for comprehensive recall.
• Cross-Platform: Integrates with Telegram, Discord, Lark (Feishu), and offers a built-in Web/CLI.
• MCP Protocol: Supports Model Context Protocol (HTTP/SSE/Stdio) for tool execution and extensibility via external servers.
• Subagents & Skills: Enables creation of specialized sub-agents and dynamic skill management for bot personalities.
• Web UI: Modern Vue 3 dashboard with real-time chat, tool visualization, and filesystem browser.
• Model Agnostic: Compatible with OpenAI-compatible, Anthropic, and Google Generative AI providers.

Maintenance & Community

• Community: An active [Telegram Group] is provided for support and discussion.
• Roadmap: A development roadmap is available.
• Development: Contribution guidelines are detailed in CONTRIBUTING.md.

Licensing & Compatibility

• License: AGPLv3.
• Compatibility: The AGPLv3 license is a strong copyleft license. It requires that any modifications or derivative works distributed must also be made available under the AGPLv3. This can impose significant restrictions on commercial use or integration into proprietary software without open-sourcing the combined system.

Limitations & Caveats

• The AGPLv3 license presents substantial considerations for commercial adoption and integration with closed-source projects.
• Specific performance benchmarks, known bugs, or detailed deprecation schedules are not detailed in the provided README snippet.
• Setup requires Docker and may necessitate root privileges for the install script.