Octop

Overview

Octop is an open-source, self-hosted AI assistant platform for households and small teams. It runs a single process serving a web dashboard, CLI, IM channels, cron automation, and APIs, with all data under ~/.octop/ on your own machine.

Key Features

• Multi-user and multi-agent operation with JWT authentication, per-user agents, and an admin role.
• Self-hosted local-first design that keeps config, chats, workspaces, and credentials on your machine.
• Web dashboard, CLI, desktop client, and HTTP/SSE/WebSocket API access.
• IM channels include Feishu, DingTalk, QQ, WeChat, Telegram, Discord, and WeCom.
• Expert library, MBTI persona templates, expert sharing, and AgentTeams beta for multi-step work.
• Knowledge base RAG, plugins, connectors with OAuth and MCP, and ACP integration for IDE workflows.

Use Cases

• Personal assistant for weekly reports, notes, scheduling, and persistent memory.
• Family sharing with one admin account and different agents or experts per member.
• Team helper that bridges IM platforms and routes tasks into group chats.
• Developer boost through ACP delegation to OpenCode or Claude Code, terminal AI, and browser automation.

Getting Started

• Install on macOS or Linux with the one-line installer, or use PowerShell or cmd installers on Windows.
• Run octop init to create the SQLite database, JWT secret, and first admin account under ~/.octop/.
• Run octop run, then open the web dashboard on local port 8088.
• For production, Docker Compose is recommended; first boot creates the admin account and writes credentials to /data/.octop/credential.txt unless OCTOP_DEFAULT_PASSWORD is set.

Deployment & Requirements

• Supported operating systems include macOS, Linux, and Windows.
• No pre-installed Python is required; the installer uses uv to provision Python 3.12 in an isolated venv under ~/.octop/.
• A modern multi-core CPU with a few GB of RAM is recommended for the process plus model and embedding caches, with enough disk for the database, agent workspaces, and document corpora.
• The control plane uses SQLite by default, with PostgreSQL optional; workspace backends can include local disk, Docker sandbox, PostgreSQL, or COS/S3.

Before You Adopt

• License: MIT. Review its terms before using, modifying, or distributing the project.