Hexabot v3 – Open-Source AI Workflow Automation Platform

TL;DR

TL;DR: Hexabot v3 is an open-source AI workflow automation platform that combines YAML-defined workflows, LLM-powered agents, MCP tool integration, and multi-channel support — deployable on your own server with a single CLI command.

Source and Accuracy Notes

• Project page: hexabot.ai
• Source repository: github.com/Hexastack/Hexabot
• License: NOASSERTION (verified via GitHub API; no standard SPDX license file found in repo — verify independently before commercial use)
• HN launch thread: news.ycombinator.com/item?id=41724497
• Latest release: v3.3.4 (verified via git ls-remote --tags)

What Is Hexabot v3?

Hexabot v3 is an open-source automation platform that brings together AI agents, workflow definitions, and conversational channels under one roof. The project describes itself as:

“Automate the Boring, Keep the Magic — Build and run agentic workflows across channels with YAML, tools, MCP, memory and RAG.”

Key characteristics verified from the README and GitHub repository:

• License: Proprietary/NOASSERTION — not a standard open-source license. Commercial use requires separate agreement. Check hexabot.ai/license before deploying in production for clients.
• Language: TypeScript/Node.js monorepo
• Latest version: v3.3.4 (as of July 2026)
• Stars: 1,121 on GitHub
• Default ports: Admin UI and API at http://localhost:3000

Setup Workflow

Prerequisites

• Node.js ^24.17.0
• One package manager: npm, pnpm, yarn, or bun
• Docker (optional, for Docker-based services)
• At least 2 GB RAM recommended

Step 1: Install the CLI

npm install -g @hexabot-ai/cli
```bash

Or run without a global install:

```bash
npx @hexabot-ai/cli --help
```bash

### Step 2: Create and run a project

```bash
hexabot create my-project
cd my-project
hexabot dev
```bash

The CLI auto-detects your package manager. To force a specific one:

```bash
hexabot create my-project --pm npm
```bash

Default local endpoints after startup:

- Admin UI: `http://localhost:3000`
- API: `http://localhost:3000/api`
- API docs (non-production): `http://localhost:3000/docs`

### Step 3: Deploy with Docker (production)

```bash
hexabot docker up
hexabot docker logs
hexabot docker ps
```bash

Useful Docker commands:

```bash
hexabot docker up --services <list>
hexabot docker down
hexabot docker start
```bash

### Step 4: Configure environment

```bash
hexabot env init      # Initialize environment variables
hexabot env list      # Show current env vars
hexabot check         # Validate configuration
hexabot config show   # Display current config
hexabot config set    # Update a config key

Deeper Analysis

Core Architecture

Hexabot v3 is structured around four pillars:

1. Agentic Workflows — YAML workflow definitions with typed runtime contracts, letting you define automation logic declaratively
2. Actions — Schema-validated input/output/settings blocks that power workflow behaviour
3. Channels — Multi-channel support (web, Discord, WhatsApp, etc.) with session continuity
4. MCP Integration — Model Context Protocol support for tool/context interoperability with external AI clients

Supported LLMs

The platform integrates with multiple LLM providers:

• Claude (Anthropic)
• GPT-4/GPT-4o (OpenAI)
• Gemini (Google)
• Mistral
• Llama (via Ollama)
• DeepSeek
• Custom endpoints via OpenAI-compatible API

Data Layer

• TypeORM as the standard backend ORM
• SQLite for local development (zero-config)
• PostgreSQL for production (first-class support)

Configure via DB_TYPE and DB_* environment variables.

MCP Integration

Hexabot v3 exposes MCP-compatible tool endpoints, allowing it to serve as a tool provider for external AI clients like Claude and Cursor. Tools are generated from the platform’s action system.

Practical Evaluation Checklist

• Node.js 24.17+ installed (node --version)
• CLI installed and hexabot --help works
• Project created with hexabot create
• Admin UI accessible at localhost:3000
• Docker deployment works (hexabot docker up)
• API docs accessible at localhost:3000/docs
• Environment variables configured for production
• PostgreSQL configured for production use
• MCP endpoint reachable (if integrating with external AI)

Security Notes

• Change default admin credentials immediately after first run
• Use hexabot check to audit configuration before production exposure
• PostgreSQL is recommended over SQLite for any internet-facing deployment
• The NOASSERTION license means you should verify commercial usage terms at hexabot.ai/license before building client projects
• Run behind a reverse proxy (nginx/Caddy) with TLS in production

FAQ

Q: Is Hexabot free to use on a VPS? A: The repository is publicly available and the core platform is open-source. However, the license is “NOASSERTION” rather than a standard SPDX license (MIT, Apache 2, GPL), so commercial deployment should be verified against hexabot.ai’s current licensing terms.

Q: What is the minimum server spec? A: The README lists no explicit minimum. For light usage, a VPS with 2 GB RAM and a single CPU core should suffice. For production with multiple channels and LLM inference, 4 GB+ RAM is recommended.

Q: Can I use it without Docker? A: Yes. The CLI-based workflow (hexabot dev / hexabot start) runs directly on Node.js. Docker is optional.

Q: Does it support MCP tools from other servers? A: Yes. Hexabot v3 has first-class MCP integration — it can both consume tools from external MCP servers and expose its own actions as MCP tools to other clients.

Q: How does it compare to n8n or Temporal? A: Unlike n8n, Hexabot is specifically designed around LLM-native agents and conversational channels rather than generic webhook/node automation. Unlike Temporal, it uses YAML-defined workflows rather than durable code execution. The MCP integration layer is also more explicit in Hexabot v3.

Conclusion

Hexabot v3 fills a specific niche: open-source AI workflow automation with first-class MCP support, multi-channel chat, and a developer-friendly CLI. Its YAML-based workflow definition makes it more accessible than Temporal for teams that want declarative automation without writing durable code. The self-hosted deployment path is straightforward — one CLI command gets you a running instance.

If you need a self-hosted alternative to platforms like Voiceflow or Manychat with LLM agents and MCP tool interoperability, Hexabot v3 is worth evaluating. Start with npx @hexabot-ai/cli create my-project on a test machine to get a feel for the workflow authoring experience before committing to a VPS deployment.

Project: hexabot.ai | Repo: github.com/Hexastack/Hexabot | v3.3.4 (July 2026)