paarths-collab/hstack

★ 13JavaScriptAudience · developerComplexity · 3/5Setup · hard

Mindmap

mindmap
  root((hstack))
    What it does
      Deploy AI agent
      Auto configure
      Server hardening
    Messaging
      Telegram
      WhatsApp
      Discord
      Slack
    Commands
      hermes-deploy
      Status check
      Update agent
      Backup
    Setup
      SSH access
      API keys
      Claude Code

mindmap root((hstack)) What it does Deploy AI agent Auto configure Server hardening Messaging Telegram WhatsApp Discord Slack Commands hermes-deploy Status check Update agent Backup Setup SSH access API keys Claude Code

Click or tap to explore — scroll the page freely

Things people build with this

USE CASE 1

Deploy your own AI assistant on a VPS that responds to your Telegram messages without any cloud lock-in

USE CASE 2

Connect an AI agent to a Slack workspace with server-level memory and a personality you define

USE CASE 3

Set up a self-hosted AI chatbot on WhatsApp with credentials stored securely on the server, not in the chat

Tech stack

JavaScriptClaude CodeBash

Getting it running

Difficulty · hard Time to first run · 30min

Requires a server with SSH access, a Claude Code installation, and API keys for both the AI model and each messaging platform you want to connect.

In plain English

hstack is a deployment toolkit that sets up a self-hosted AI agent on your own server and connects it to your phone via Telegram, WhatsApp, Discord, or Slack. The goal is to let someone with an empty server go from nothing to a working AI assistant in about 30 minutes by pasting a single command into Claude Code (Anthropic's coding assistant). Claude Code then handles the installation, configuration, and server hardening automatically, stopping only to ask for the credentials it cannot generate itself. The underlying agent software being installed is called Hermes Agent. Installing Hermes by hand involves a long series of steps, and many things can break: environment PATH issues, memory leaks in the gateway process, stale lock files, and configuration quirks specific to each messaging platform. hstack pre-solves all of those known issues and pins a specific tested version of Hermes so you do not land on a version that has new bugs. The toolkit is built as a set of slash commands that Claude Code can run. The main one, /hermes-deploy, covers everything end to end: installing Hermes on the server (locally or over SSH), configuring the AI model and API key, connecting your messaging platforms, setting up skills and memory, giving the agent a personality, and verifying that it is running. Separate smaller commands handle things like restarting the agent, checking its status, applying updates, and taking backups. There are 19 commands in total. For messaging platform setup, Claude walks you through each credential it needs step by step, from creating a Telegram bot with BotFather to scanning a WhatsApp QR code. Secrets are stored in a protected file on the server rather than pasted into the chat. Security defaults include restricting the agent to an allowlist of authorized users and binding services to localhost rather than exposing them directly to the internet. The README notes that most of the work is done by Claude Code once it has server access, and that the project productizes a setup process the author originally spent four hours doing by hand.

Copy-paste prompts

Prompt 1

Using hstack's /hermes-deploy command, walk me through deploying a Hermes AI agent on my Ubuntu VPS and connecting it to Telegram step by step

Prompt 2

Help me configure hstack to connect my Hermes agent to both WhatsApp and Discord, and show me how credentials are stored securely on the server

Prompt 3

Show me how to use hstack's backup and update slash commands to maintain a Hermes agent I already have running

Prompt 4

Help me troubleshoot a hstack deployment where the gateway process has a memory leak or a stale lock file is blocking the install

Open on GitHub → Explain another repo

← paarths-collab on gitmyhub — every repo by this author, as a profile.

Verify against the repo before relying on details.