AutoBot

AI-powered autonomous platform for Linux administration, multi-modal AI interaction, and intelligent workflow automation.

AutoBot is a self-hosted platform that combines a conversational AI assistant with a distributed automation engine. Chat with AI models, run browser and terminal automation, manage knowledge bases, analyze codebases, and administer a fleet of Linux nodes — all from a single web dashboard.

Status: Active development · v1.5.0

---

Screenshots

AI Chat with voice, terminal, browser & desktop tabs	Visual Workflow Builder with execution strategy selection
Real-time codebase analytics and code quality scoring	Service Lifecycle Manager — fleet health across all nodes

---

What You Can Do

Module	What it does
Chat	Conversational AI backed by Ollama and 8 LLM provider types. Voice input/output, file attachments, inline terminal, embedded browser, and VNC desktop — all within a single chat session
Knowledge Base	Upload documents and code, then query via semantic search or RAG-enhanced AI answers. Browse by category or explore the entity knowledge graph
Automation	Visual workflow builder with sequential, parallel, pipeline, collaborative, and adaptive execution strategies. Describe workflows in natural language
Analytics	Index and analyze codebases in real-time. Code quality scores, duplicate detection, API coverage, hardcoded value scanning, and business intelligence dashboards
Vision	AI screen capture and element analysis. Identify UI components, extract text from screenshots, and feed results into automation workflows
SLM Admin	Service Lifecycle Manager — deploy, monitor, update, and roll back services across a fleet of Linux nodes via Ansible automation

---

Requirements

Component	Minimum	Recommended
OS	Linux or WSL2	Ubuntu 22.04 LTS
RAM	16 GB	32 GB+
CPU	8 cores	16+ cores
Storage	50 GB	200 GB+ (model + knowledge base storage)
GPU	—	NVIDIA (CUDA) or Intel NPU for accelerated inference
Python	3.10	3.12 (conda env)
Node.js	18	20

---

Quick Start

1. Clone the repository

git clone https://github.com/mrveiss/AutoBot-AI.git
cd AutoBot-AI

2. Run initial setup

bash setup.sh

Installs dependencies, configures environment variables, initializes Redis databases, and prepares all containers.

3. Start AutoBot

# Development mode — auto-reload, debug logging
bash run_autobot.sh --dev

# Production mode
bash run_autobot.sh --prod

# Skip container rebuild if images already exist
bash run_autobot.sh --dev --no-build

4. Open the interface

Interface	URL	Default credentials
User Frontend	https://172.16.168.21	admin / admin
Backend API docs	https://172.16.168.20:8443/docs	—
SLM Fleet Admin	https://172.16.168.19	admin / admin
VNC Desktop	http://127.0.0.1:6080	—

Change the default password on first login.

5. Verify services are running

scripts/start-services.sh status

---

How to Use

Chat

The main interface at https://172.16.168.21/chat. Start a conversation and switch between:

• Chat — send messages to the AI assistant
• Files — browse and manage files on the server
• Terminal — run shell commands without leaving the chat
• Browser — embedded Playwright browser with visual automation
• noVNC — full remote desktop access

Input toolbar shortcuts: KB searches the knowledge base, Explain, Summarize, and Translate are one-click AI actions. Mic button enables voice input. Attach files with the paperclip.

---

Knowledge Base

Go to Knowledge → Search. Two modes:

• Traditional Search — fast semantic search across all indexed documents
• RAG Enhanced — full AI pipeline with retrieval-augmented generation for richer answers

Filter by category or access level (Platform, Public, System, User). Upload and manage documents at Knowledge → Manage. Explore entity relationships at Knowledge → Graph.

---

Automation

Go to Automation → Workflow Builder. Choose an execution strategy:

Strategy	When to use
Sequential	Tasks with strict dependencies
Parallel	Independent tasks that can run concurrently
Pipeline	Output of one stage feeds into the next
Collaborative	Multiple agents sharing context in real-time
Adaptive	Strategy adjusts dynamically based on progress

Use Natural Language mode to describe what you want in plain English — AutoBot generates the workflow automatically. Run workflows with Runner and review history under History.

---

Analytics

Go to Analytics → Codebase. Point it at a codebase directory and click Index Codebase. Available analyses:

• Code quality score, code smells, health score
• Duplicate detection, hardcoded value scanning
• API endpoint coverage, NPU compatibility check
• Communication patterns, performance metrics
• Export to report

---

Vision

Go to Vision → Image Processing. Click Capture & Analyze to take a screenshot of the connected desktop and run AI element detection. Enable Auto-refresh to continuously monitor a screen. Use detected elements as inputs for browser automation workflows.

---

SLM Fleet Admin

Accessed at https://172.16.168.19 or via the SLM Admin button in the top-right corner.

Section	Purpose
Fleet Overview	Real-time CPU/RAM/disk health across all nodes
Orchestration	Start, stop, restart services on any node
Deployments	Roll out code updates via Ansible playbooks
Backups	Schedule and restore node backups
Replication	Manage data replication between nodes
Code Sync	Push code updates to the fleet from git
Agents	View and manage AI agents running across nodes
Updates	Check and apply software updates
Skills	Deploy AI skill libraries to nodes
Monitoring	Prometheus metrics and alerting

---

Architecture

AutoBot runs as a distributed system. Each component is isolated on its own node:

                        User Browser
                             │
                    ┌────────▼─────────┐
                    │  Frontend (.21)  │
                    │  Vue 3 + Vite    │
                    └────────┬─────────┘
                             │ HTTPS / WebSocket
                    ┌────────▼─────────┐
                    │  Backend (.20)   │
                    │  FastAPI · Py3.12│
                    │  RTX 4070 · 22c  │
                    └──┬──┬──┬──┬──┬──┘
                       │  │  │  │  │
           ┌───────────┘  │  │  │  └────────────┐
           │              │  │  │               │
     Redis (.23)    AI Stack  Browser VM    NPU VM (.22)
     Vector DB +    (.24)     (.25)         Intel OpenVINO
     Session store  LLM       Playwright    Hardware accel
                              automation
                                   │
                          ┌────────▼────────┐
                          │   SLM (.19)     │
                          │  Fleet Manager  │
                          │  Ansible ctrl   │
                          └─────────────────┘

Tech stack at a glance:

• Backend: Python 3.12, FastAPI, asyncpg, ChromaDB, Ollama
• Frontend: Vue 3, TypeScript, Vite
• Data: Redis Stack, PostgreSQL (user management), ChromaDB (vectors)
• Infra: Ansible, systemd, nginx, self-signed TLS

---

Service Management

# Start / stop / restart all services
scripts/start-services.sh start
scripts/start-services.sh stop
scripts/start-services.sh restart

# Start specific service
scripts/start-services.sh start backend

# View logs
scripts/start-services.sh logs backend
scripts/start-services.sh logs frontend

# Open SLM GUI
scripts/start-services.sh gui

Direct systemctl:

sudo systemctl start autobot-backend
sudo systemctl status autobot-backend
journalctl -u autobot-backend -f

---

Documentation

Topic	Link
Developer Reference	docs/developer/AUTOBOT_REFERENCE.md
Getting Started (full)	docs/GETTING_STARTED_COMPLETE.md
API Documentation	docs/api/COMPREHENSIVE_API_DOCUMENTATION.md
Redis Architecture	docs/api/redis-documentation.md
Distributed Architecture	docs/architecture/DISTRIBUTED_6VM_ARCHITECTURE.md
Service Management	docs/developer/SERVICE_MANAGEMENT.md
Troubleshooting	docs/troubleshooting/INDEX.md
Infrastructure & Scripts	autobot-infrastructure/README.md

For development workflow, coding standards, and rules: see CLAUDE.md.

---

Copyright © 2025 mrveiss. All rights reserved.