nanobot
HKUDS/nanobot
An ultra-lightweight personal agent framework from the University of Hong Kong's Data Intelligence Lab, providing memory, tools, MCP and chat-app integrations in a small amount of Python.
Why pick itFar less code than OpenClaw, so it's easy to read and modify. Good for self-hosting or learning agent architecture; built by a Hong Kong team.
Install
- GitHub
- HKUDS/nanobot
- Website
- nanobot.wiki
- Stars
- ★ 49.3k
- License
- MIT
- Language
- Python
- Last push
- Oct 25, 2026
- Category
- Rising in 2026
Good for
- Use a personal agent from a browser WebUI, the terminal, or chat apps like Telegram, Discord, Slack and WeChat
- Read PDF, Word, Excel and PowerPoint files and summarize reports
- Run scheduled automations and long-horizon goals with session history and long-term memory
- Study its small Python codebase to learn agent architecture or extend it
Getting started
- Have Python 3.11+ and a supported model provider or local model server ready
- Run pip install nanobot-ai && nanobot onboard to set it up
- Send a first message in the WebUI or terminal, then connect chat apps or MCP following the docs
Before you install
nanobot does not supply a model account; model fees or local hardware needs depend on the service you choose.
Summarized by AI from the project docs; check the official docs before relying on it.
Similar tools
- OpenClawRising in 2026 · ★ 393k
A self-hosted personal AI assistant you command through chat apps like WhatsApp, Telegram and Discord; it can operate your computer and browser and run scheduled tasks.
- Matt Pocock's SkillsRising in 2026 · ★ 303k
TypeScript educator Matt Pocock's personal agent skills, covering engineering workflows like writing PRDs, splitting issues, TDD, refactoring and code review.
- DeepSeek HarnessRising in 2026 · ★ 261k
An open-source agent framework DeepSeek released in August 2026. Models, tools, skills, sandboxes and UIs are all composed as plugins; it launches a local Web UI by default.
- Hermes AgentRising in 2026 · ★ 257k
Nous Research's always-on agent that builds memory from use and generates its own skills; works in the terminal and many chat platforms.
- claude-memRising in 2026 · ★ 103k
Automatically records what an agent did in each session, compresses it into AI summaries, and injects relevant context into your next conversation.
Source GitHub · Updated Oct 24, 2026 · All rights belong to the original creators