Claude Code Club
Agents

Own a 24/7 AI Assistant With These 10 Free Repos (OpenClaw, Hermes Agent, Ollama and More)

11 minute readUpdated October 2026Explore more

TL;DR

You can run your own AI assistant that keeps working while you sleep. These 10 free GitHub repos cover every piece. OpenClaw and Hermes Agent are the assistants themselves: they live in your chat apps and run tasks on a schedule. Ollama and Open WebUI let you run open AI models on your own machine with a ChatGPT style app. AionUi and cc-haha are desktop apps that run Claude Code and other agents side by side. Google Workspace CLI, Activepieces and notebooklm-py connect your assistant to email, calendar, everyday apps and research. Claude Code UI lets you check in from your phone. Each section has the GitHub link, live star count, install steps from the repo's own README, and the honest safety catch. Read the safety notes before you give any assistant access to your email or files.

Learn Claude Code. Earn income. Only $9.

👉 https://www.skool.com/claudecodeclub

The 10 repos to own a 24/7 AI assistant

Star counts below were pulled from GitHub on October 2, 2026, so they will be higher by the time you read this. The list covers four jobs: the assistant itself, running AI models on your own machine, connecting it to your email, calendar and apps, and checking in from anywhere. You don't need all 10. Pick one assistant, then add the pieces you actually need.

1. OpenClaw: a personal AI that lives in your chat apps

openclaw/openclaw

391,191 stars on October 2, 2026. MIT license.

What it does: an open-source AI assistant that runs on your own computer and talks to you in the apps you already use, like WhatsApp, Telegram, Slack, Discord and iMessage, plus native apps for macOS, iOS, Android, Windows and Linux. Claude, Codex and local models plug in as swappable models. Its README says state, memory and credentials live on your hardware.

How to install it (macOS, Linux or WSL2): the installer sets up Node.js if needed and starts the setup wizard automatically.

bashcurl -fsSL https://openclaw.ai/install.sh | bash

Already have Node.js? Install the package instead, then run the setup wizard yourself:

bashnpm install -g openclaw@latest --allow-scripts=openclaw
openclaw onboard --install-daemon

After setup, check it is running and open the control panel. Send it a message there to confirm it works:

bashopenclaw gateway status
openclaw dashboard

2. Hermes Agent: an assistant that learns and works on a schedule

NousResearch/hermes-agent

250,770 stars on October 2, 2026. MIT license. Made by Nous Research.

What it does: a self-improving agent from Nous Research. It creates skills from experience, remembers past conversations, and builds a picture of who you are across sessions. It has a built-in scheduler for daily reports and other unattended jobs, and you can talk to it from Telegram, Discord, Slack, WhatsApp, Signal or the terminal. The README says it can run on a $5 VPS, so it does not have to live on your laptop.

How to install it (macOS, Linux or WSL2):

bashcurl -fsSL https://hermes-agent.nousresearch.com/install.sh | bash

Then reload your shell and start chatting. `hermes model` picks your AI provider:

bashsource ~/.zshrc
hermes model
hermes

To talk to it from your phone, run `hermes gateway setup` and then `hermes gateway start`, and message the bot.

3. Ollama: run open AI models on your own machine

ollama/ollama

182,067 stars on October 2, 2026. MIT license.

What it does: downloads and runs open AI models on your own computer, so chats can stay private and work offline. It can also connect to agents like Claude Code and OpenClaw.

How to install it (Mac or Linux):

bashcurl -fsSL https://ollama.com/install.sh | sh

Then chat with a model, or connect it to OpenClaw as a personal assistant:

bashollama run gemma4
ollama launch openclaw

4. Open WebUI: your own ChatGPT style app

open-webui/open-webui

153,812 stars on October 2, 2026. Open WebUI License (custom, with a branding requirement).

What it does: a friendly chat app in your browser for your own models. Point it at Ollama for local models, or at an OpenAI API key.

How to install it: if Ollama is already on your computer and you have Docker, run this, then open http://localhost:3000:

bashdocker run -d -p 3000:8080 --add-host=host.docker.internal:host-gateway -v open-webui:/app/backend/data --name open-webui --restart always ghcr.io/open-webui/open-webui:main

Or install with pip. The README says to use Python 3.11 to avoid problems. It then runs at http://localhost:8080:

bashpip install open-webui
open-webui serve

5. AionUi: coworkers that never sleep

iOfficeAI/AionUi

33,289 stars on October 2, 2026. Apache 2.0 license.

What it does: a free desktop app with a built-in AI agent that reads and writes files, searches the web and runs multi-step tasks. It auto-detects Claude Code, Codex, Hermes Agent, OpenClaw and others and runs them side by side. Scheduled tasks let it work on its own, and a WebUI mode or Telegram lets you check in from your phone.

How to install it: download the installer for your system from the Releases page (https://github.com/iOfficeAI/AionUi/releases), or on a Mac:

bashbrew install aionui

Then open the app, paste any AI API key, and start a task.

6. Google Workspace CLI: let your assistant handle Gmail and Calendar

googleworkspace/cli

31,236 stars on October 2, 2026. Apache 2.0 license.

What it does: one command-line tool (`gws`) for Gmail, Drive, Calendar, Docs, Sheets and the rest of Google Workspace, with clean JSON output that AI agents can read. It ships 100+ agent skills, including ready-made recipes for Gmail, Drive, Docs, Calendar and Sheets.

How to install it: you need Node.js 18+ and a Google Cloud project (the setup command can create one for you if you have the gcloud CLI).

bashnpm install -g @googleworkspace/cli
gws auth setup
gws auth login

Then add the skills to your agent:

bashnpx skills add https://github.com/googleworkspace/cli

7. Activepieces: plug your assistant into the apps you already use

activepieces/activepieces

24,860 stars on October 2, 2026. MIT license for the Community Edition. Enterprise features use a commercial license.

What it does: an open-source replacement for Zapier. You build automations in a no-code editor, and its README says its 280+ pieces (app integrations) are also available as MCP servers you can use with Claude Desktop, Cursor or Windsurf.

How to install it: the README does not list a one-line install. Self-hosting steps live in the official deploy guide: https://www.activepieces.com/docs/install/overview. Or ask Claude Code:

promptRead the Activepieces deploy guide at https://www.activepieces.com/docs/install/overview and walk me through the simplest self-hosted setup on this computer, one step at a time. Explain each command before running it.

8. notebooklm-py: research on autopilot

teng-lin/notebooklm-py

19,581 stars on October 2, 2026. MIT license.

What it does: gives Claude Code and other agents full control of Google's NotebookLM (now also called Gemini Notebook). Your assistant can add sources, ask cited questions, and generate podcasts, slide decks, quizzes and reports from your documents, so Gemini does the heavy reading instead of your Claude tokens.

How to install it: the README recommends uv (or pipx). The first login downloads a browser (about 170 MB) and opens a Google sign-in.

bashuv tool install "notebooklm-py[browser]"
notebooklm login
notebooklm auth check --test --json

Then install the skill so Claude Code knows how to use it:

bashnotebooklm skill install

9. cc-haha: a desktop for your Claude Code agents

NanmiCoder/cc-haha

14,841 stars on October 2, 2026. MIT license. Community project, not made by Anthropic.

What it does: a desktop workspace for Claude Code on macOS, Windows and Linux. You get multiple sessions in tabs, file-by-file diff review, Git worktrees, scheduled tasks, Agent Teams, chat app control (Telegram, Slack, WhatsApp and more), and phone access by scanning a QR code. It works with Claude, ChatGPT, Grok or local models through Ollama.

How to install it: download the installer for your system from the Releases page (https://github.com/NanmiCoder/cc-haha/releases). On first launch, set your model provider, API key and default model in Settings.

10. Claude Code UI: run Claude Code from your phone or browser

siteboon/claudecodeui

13,907 stars on October 2, 2026. AGPL 3.0 license. Now also called CloudCLI.

What it does: a web app for Claude Code, Cursor CLI and Codex that works on desktop, tablet and phone. It finds all your existing Claude Code sessions automatically and adds a chat, file explorer, Git tools and a terminal.

How to install it: you need Node.js 22 or newer.

bashnpx @cloudcli-ai/cloudcli

Then open http://localhost:3001. To run agents in an isolated sandbox instead (experimental, needs Docker's `sbx` CLI):

bashnpx @cloudcli-ai/cloudcli@latest sandbox ~/my-project

Quick start

  1. 1Pick a spare computer, virtual machine or cheap server for your assistant. Do not start on the laptop with all your files and logins.
  2. 2Install one assistant: OpenClaw if you live in chat apps, Hermes Agent if you want it to learn and run scheduled jobs.
  3. 3Connect one tool with the least access possible, like Google Workspace CLI with only Calendar, and leave approvals on.
  4. 4Add Ollama later if you want private, offline chats, and Claude Code UI if you want to check in from your phone on your home network.
Learn Claude Code. Earn income. Only $9.

👉 https://www.skool.com/claudecodeclub

Common questions

  • Are these 10 repos free?

    The software is free and open source. You still pay for the AI models you connect, like a Claude or OpenAI API key, unless you run local models with Ollama. Some have optional paid extras: Nous Portal for Hermes Agent, CloudCLI Cloud for Claude Code UI, and enterprise features in Activepieces.

  • Is it safe to give an AI assistant my email and files?

    Only with care. Emails and web pages can hide instructions that trick the assistant into sending, deleting or leaking things. Run it on a separate machine or sandbox, give it the least access it needs, keep approvals on for anything that sends or deletes, and never paste API keys or passwords into a chat with it.

  • Which one should I install first?

    One assistant: OpenClaw or Hermes Agent. Both run on a schedule and answer you in chat apps. Add the other repos only when you hit a real need, like email (Google Workspace CLI) or phone access (Claude Code UI).

  • Do I need a powerful computer?

    Not if you use cloud models like Claude through an API key. You only need strong hardware (lots of memory, ideally a good graphics chip) if you run local models with Ollama, and small local models are weaker than Claude.

  • Does it keep working when my laptop is closed?

    Self-hosted assistants only run while the machine running them is on and awake. That is why many people put the assistant on an always-on spare computer or a small server. Hermes Agent's README says it can run on a $5 VPS.

Keep going

Get more repos like these inside Claude Code Club for $9/mo

Get 650+ plug-and-play skills, MCPs & prompts, plus 8,000+ members - $9/mo, cancel anytime.

Join the Club