TencentCloud/Octop

A smarter, self-hosted AI assistant — multi-user, multi-agent.

What it solves

Octop is a self-hosted AI assistant platform designed for individuals, families, and small teams. It addresses the need for a private, multi-user AI environment where users can maintain their own specialized agents and data locally on their own machine, avoiding vendor lock-in and privacy compromises.

How it works

Octop runs as a single process that integrates a web dashboard, a command-line interface (CLI), and various instant messaging (IM) channels. It is built on the "Harness" stack, which provides the agent runtime, a multi-platform IM bridge, hierarchical memory, and browser automation. All user data, including conversations, workspaces, and credentials, is stored in a local SQLite database and directory structure under ~/.octop/.

Who it’s for

It is intended for users who want a fully self-hosted AI assistant with multi-user support, those who need to integrate AI into their existing communication tools (like Discord or Feishu), and developers who want to extend AI capabilities via the Agent Client Protocol (ACP) for IDE integration.

Highlights

  • Multi-user Expert Team: Supports an admin and shared household with a built-in expert library to switch specialists per scenario.
  • Extensive IM Integration: Connects with Feishu, DingTalk, QQ, Discord, and WeCom.
  • Local-First Security: Features JWT multi-user isolation, shell command guardrails, and PII redaction to keep data local.
  • Advanced AI Capabilities: Includes headless Chromium sessions for web automation, remote desktop access for GUI apps, and an interactive shell in the browser.
  • ACP Bidirectional Support: Allows external tools to use Octop agents or delegates tasks to external coding agents like Claude Code or OpenCode.
  • Flexible Backends: Supports local disk, Docker containers, PostgreSQL, and S3/COS for workspace storage.

Related

  • Project
    octos-org/octos

    An embeddable AI agent harness kernel written in Rust that provides the execution loop, memory, and coordination infrastructure for agent-powered applications.

  • Project
    kgoedecke/doop

    Doop is an open‑source, real‑time design canvas where humans and AI agents (via Claude Code or a built‑in Doop Agent) edit HTML frames together. It offers live multiplayer features, AI‑driven design tasks, private sharing, and can be self‑hosted with a single Docker command or `bun run dev`. The AI backend works with Anthropic (free tier) or user‑provided OpenAI/ChatGPT keys.

  • Project
    ThinkInAIXYZ/deepchat

    An open-source, local-first AI agent desktop client that integrates multiple LLM providers, MCP/ACP protocols, and remote control capabilities for messaging apps.

  • Project
    modelscope/ms-agent

    A lightweight framework for building autonomous agents capable of deep research, code generation, and multimodal tasks using a knowledge-driven skill system.

  • Project
    EpicStaff/EpicStaff

    A self-hosted AI agent orchestration platform that allows operations teams to visually design, manage, and audit agent workflows while engineers handle the system integrations.