VeryBot – Self-Hosted AI Assistant for Work

TL;DR

TL;DR: VeryBot is a self-hosted AI assistant platform you run via npm install -g verybot. It covers chat, task management, team workflows, and channel integrations — all under your own control.

What Is VeryBot?

Most AI assistants run in the cloud. VeryBot runs on your own machine. It is an open-source, self-hosted work assistant that unifies chat, task tracking, team collaboration, and automation into a single UI — with support for connecting external channels and MCP-powered tools.

Key capabilities:

• Chat interface with an AI model of your choice (OpenAI, Anthropic, local Codex/Claude CLI)
• Task management — create, assign, and track tasks from within conversations
• Team workspaces — shared context and workflows for small teams
• Playbooks — save proven workflows as reusable templates
• Channel integrations — Telegram, Discord, Slack, and WhatsApp
• MCP support — connect custom tools and skills without modifying core code
• Local-first storage — runtime data under ~/.verybot, nothing leaves your server

Source and Accuracy Notes

⚠️ This section is MANDATORY. All links must be verified from actual source, not guessed.

• Project page: verybot.io — verified via README navigation link
• Source repository: github.com/charlie0077/verybot — README read end-to-end
• License: MIT (verified via https://opensource.org/license/mit)
• HN launch thread: news.ycombinator.com/item?id=47176282

Setup Workflow

Step 1: Install Node.js and verybot

VeryBot requires Node.js 22 or later and npm. Install it with:

npm install -g verybot
verybot

The Control UI starts at http://localhost:28789. On first launch, VeryBot prints a gateway token in the terminal output (GATEWAY_TOKEN: ...). Copy it from there, or retrieve it later with:

verybot config get GATEWAY_TOKEN

Step 2: Configure Your Model

Open the Control UI and go to Settings -> Agent to choose a model provider. Two options:

1. API key provider — enter your OpenAI or Anthropic key
2. Local CLI — if codex or claude CLI is already installed and working on the machine, no key is needed

Step 3: Connect External Channels

VeryBot can receive and send messages through workplace chat platforms. Supported channels:

• Telegram
• Discord
• Slack
• WhatsApp

Configure each channel through the Control UI under Settings -> Channels. Once connected, you can interact with your assistant from any of these platforms.

Step 4: Use Playbooks

Playbooks are saved workflows you can replay at any time. If your team has a repeatable process — a weekly report, a deployment checklist, an onboarding sequence — save it as a playbook and trigger it with a single prompt.

Step 5: Extend with MCP

VeryBot supports the Model Context Protocol. Add custom tools and skills without modifying the core application. MCP integrations are configured in Settings -> MCP.

Step 6: Access from LAN or Internet

By default, the Control UI binds to localhost. To access it from another machine on your network:

verybot --host 0.0.0.0

For internet access, point a domain at your machine and configure TLS termination at your router or reverse proxy.

Deeper Analysis

Local-First Data

All runtime data lives under ~/.verybot. Nothing is sent to a third-party cloud unless you explicitly configure a cloud model provider. This makes VeryBot suitable for teams that handle sensitive data or operate under data-residency requirements.

Channel Integration Model

Unlike a simple Slack bot, VeryBot acts as a unified gateway — the same assistant context is available across all connected channels. This means a task created via Telegram is visible in Discord and in the web UI simultaneously.

Comparison with Cloud Assistants

| | VeryBot | Cloud AI assistants | |---|---|---| | Data storage | Local (~/.verybot) | Provider’s servers | | Channel integrations | Self-hosted (Telegram, Discord, etc.) | Provider-controlled | | Deployment | Your own machine/VPS | Managed by vendor | | Cost | Infrastructure only | Subscription + API costs | | Extensibility | MCP, custom tools | Plugin marketplace |

Practical Evaluation Checklist

• [ ] Node.js 22+ installed
• [ ] npm install -g verybot succeeds
• [ ] Control UI loads at localhost:28789
• [ ] Gateway token found and UI accessible
• [ ] Model provider configured (API key or local CLI)
• [ ] At least one channel connected
• [ ] A task created and tracked end-to-end
• [ ] Playbook saved and replayed
• [ ] LAN access configured (if needed)

Security Notes

• The gateway token authenticates the Control UI — treat it like a password
• Default binding is localhost only; expose externally only behind a TLS-terminating proxy
• Data at rest in ~/.verybot is not encrypted by default; apply filesystem-level encryption if needed
• API keys for model providers are stored in the local config; secure the config file appropriately

FAQ

Q: Who is VeryBot designed for? A: Solo operators who want a self-hosted work copilot, small teams that need shared AI context across chat platforms, and developers who want a self-hosted runtime with tools and API access.

Q: How does it compare to using Claude or ChatGPT directly? A: VeryBot layers workflow automation, task tracking, team context, and channel integrations on top of whatever model you choose. The assistant persists across sessions and can automate recurring work. Cloud assistants are session-based and do not integrate into your team’s tooling by default.

Q: How do I upgrade to a new version? A: Run npm install -g verybot@latest to pull the latest release.

Q: What if I lose the gateway token? A: Run verybot config get GATEWAY_TOKEN from the terminal to retrieve it.

Conclusion

VeryBot fills the gap between a personal AI chat tool and a full team workflow platform — entirely under your own infrastructure. If you want persistent assistant context, task tracking, and channel integrations without sending data to a third-party cloud, it is worth a look. Installation takes under five minutes with Node.js 22 already available.