nanobot
Self-hosted personal AI agent in Python that works through a browser, terminal, or chat apps.
- Stars
- 49.8k
- Forks
- 8.7k
- Licence
- MIT
- Language
- Python
- Last update
- 4 days ago
- Type
- App
What it is
nanobot is an open-source personal AI assistant you run yourself. You can use it in a browser, a terminal, or chat apps like Telegram or Discord to work with files, remember preferences, and schedule tasks. It does not include a model; you connect it to a supported model provider or a local model server.
- Summarize a report stored as a PDF, Word, Excel, or PowerPoint file
- Chat with the assistant from Telegram, Discord, or Slack while it keeps conversation history
- Schedule recurring automated tasks
- Delegate parts of a job to multiple agents
- Call it from Python code or through an OpenAI-compatible HTTP API
Great for
- People who want a self-hosted personal assistant they can inspect and customize
- Developers who want a small Python agent framework with memory, tools, and MCP support
- Users who want an assistant reachable from chat apps like Telegram, Discord, or Slack
- People who want to summarize PDF, Word, Excel, or PowerPoint files with an AI assistant
Not ideal for
- People who want a ready-made hosted service with no setup, since you must supply a model account or local model server
- Non-technical users who are uncomfortable with terminals or configuration, unless they follow the separate non-technical guide
- Teams needing a documented enterprise security model, since the README does not describe one in detail
How to get it
- 1
Install with a package manager
You'll need: Python 3.11 or newer · Access to a supported model provider or local model server
Install stable release with uv
uv tool install nanobot-aiInstall stable release with pip (inside an activated virtual environment)
python -m pip install nanobot-ai - 2
Run it
- Install the package with one of the commands above
- Run nanobot webui for browser setup, or nanobot onboard --wizard for terminal setup
- Connect a supported model provider or local model server
- Send your first message in the WebUI or terminal
- 3
Get the most out of it
- Connect a chat app such as Telegram or Discord using the chat app guide
- Configure fallback models and providers through the configuration docs
- Use scheduled automations for recurring tasks
- Add MCP integrations to extend the tools available to the agent
- Use the Python SDK or OpenAI-compatible API to build your own integrations
Downloads
Strengths
- Self-hosted, and the README says you can inspect and extend the code
- Works across a browser UI, terminal, and many chat channels
- Supports OpenAI-compatible APIs and local models, so you can choose your provider
Trade-offs
- Requires a model provider account or local model hardware, and costs depend on that choice
- The README is truncated and describes current source, which may differ from the stable release
- Setup involves Python and terminal steps for most users
How it compares
| Project | Stars | Type | Level | Updated | Best for |
|---|---|---|---|---|---|
| nanobot (this) | 49.8k | App | intermediate | 4 days ago | People who want a self-hosted personal assistant they can inspect and customize |
| langchain | 149k | Framework | intermediate | 4 days ago | Python developers building LLM-powered applications |
| CowAgent | 47.8k | App | intermediate | 4 days ago | People who want a self-hosted personal assistant reachable from their everyday chat apps |
| paperclip | 110k | App | advanced | 4 days ago | Developers already running several AI coding agents who want one place to track them |
| invisible_dots | 32.4k | AI model / agent | intermediate | 6 days ago | Developers who want a self-hosted agent and are comfortable with a command-line setup |
| Horizon | 11.6k | App | intermediate | 4 days ago | People who want a personalized daily news digest they run themselves |
Stats synced from GitHub. The assessment above was drafted with AI from the project's README on 2026-10-29 and may miss recent changes — check the README before relying on a command. MIT licence: free to use, modify and distribute, including commercially, as long as you keep the copyright notice.