ClawHub Desktop

Autonomous AI Operating System

A next-generation AI workspace that unifies multi-provider inference, self-improving agents, context memory, scheduled automation, research pipelines, and 44 power tools into a single desktop application.

Getting Started · Features · AI Providers · Architecture · API Reference

---

Getting Started

Prerequisites

Requirement	Version	Purpose
Node.js	v18+	Runtime environment
npm	v9+	Package manager
Python 3	3.8+	PDF extraction (PyPDF2)
Git	v2+	Version control

One-Command Install (Windows PowerShell)

irm https://raw.githubusercontent.com/rachidSabah/INFOHASCLAWHUB/main/install.bat -OutFile install.bat; .\install.bat

The script automatically installs Node.js (via winget), clones the repo, installs dependencies, sets up the SQLite database, and launches the dashboard at http://localhost:3000.

One-Command Install (WSL / Linux)

curl -fsSL https://raw.githubusercontent.com/rachidSabah/INFOHASCLAWHUB/main/install.sh | bash

The script installs Node.js (via nvm if missing), clones the repo, installs dependencies, runs Prisma migrations, and starts the dev server.

Manual Install

git clone https://github.com/rachidSabah/INFOHASCLAWHUB.git
cd INFOHASCLAWHUB
npm install
npx prisma db push --accept-data-loss
npx prisma generate
npm run dev

Starting the Dashboard

After installation completes, the dashboard starts automatically. If you need to start it manually:

# Development mode (recommended)
cd INFOHASCLAWHUB
npm run dev

# Production mode (faster, more stable)
npm run build
npx next start

Then open your browser to http://localhost:3000

LAN access from other devices:

npm run dev:lan    # Binds to 0.0.0.0:3000 for network access

WSL users accessing from Windows browser:

hostname -I        # Get your WSL IP, then browse to http://<WSL-IP>:3000

Uninstall

# WSL / Linux
curl -fsSL https://raw.githubusercontent.com/rachidSabah/INFOHASCLAWHUB/main/uninstall.sh | bash

# Windows (PowerShell)
irm https://raw.githubusercontent.com/rachidSabah/INFOHASCLAWHUB/main/uninstall.bat -OutFile uninstall.bat; .\uninstall.bat

---

AI Provider Setup

ClawHub supports multiple AI providers. You can configure them in Settings > Providers or via environment variables.

Gemini CLI (Default — Free, No API Key Needed)

Gemini CLI provides free access to Google's Gemini models through your Google account. No API key or credit card required.

Install on Windows:

npm install -g @google/gemini-cli

Install on WSL / Linux:

npm install -g @google/gemini-cli

Authenticate:

gemini    # Follow the Google OAuth login flow in your browser

Verify installation:

gemini --version

Once authenticated, Gemini CLI is automatically detected by ClawHub. No further configuration needed — just select a Gemini model from the model dropdown in the top bar.

Troubleshooting: If gemini is not recognized after installation, restart your terminal/WSL session to refresh the PATH. On Windows, ensure %APPDATA%\npm is in your system PATH.

DeepSeek

1. Create an account at platform.deepseek.com
2. Generate an API key from the dashboard
3. In ClawHub, go to Settings > Providers > Add Provider
4. Name: DeepSeek, Base URL: https://api.deepseek.com, paste your API key

BigModel / GLM (Z.AI)

1. Create an account at open.bigmodel.cn
2. Generate an API key
3. In ClawHub, go to Settings > Providers > Add Provider
4. Name: BigModel, Base URL: https://open.bigmodel.cn/api/paas/v4, paste your API key

OpenAI / Anthropic / Groq

1. Obtain an API key from the respective provider
2. In ClawHub, go to Settings > Providers > Add Provider
3. Enter the provider name, base URL, and API key

Proxima (Local Browser Gateway)

1. Install and start Proxima locally on port 3210
2. Automatically detected — no key needed

Hybrid Router

The Hybrid Router benchmarks all configured providers in real time, scoring them on latency, cost, and quality. It auto-selects the best provider for each request or races multiple providers and returns the fastest response. Enable it from the Power Tools menu.

---

Features

AI Inference and Chat

Feature	Description
Multi-Provider AI	Gemini CLI, DeepSeek, BigModel/GLM, OpenAI, Anthropic, Groq, Proxima, and any OpenAI-compatible endpoint
OpenAI-Compatible Gateway	Full /v1/chat/completions and /v1/models endpoints for external integrations
Real-Time Streaming	Server-Sent Events (SSE) streaming with typing indicators and progressive token delivery
Persistent Chat History	SQLite-backed storage — all conversations are saved locally and fully searchable
Multi-Tab Chat	Open and switch between multiple conversations like browser tabs
Favorites and Rename	Star conversations, rename them, and filter by starred or date
Smart Sidebar	Conversations grouped by Today/Yesterday/Date with context-menu actions
File Attachments	Attach files to any message for AI processing
Markdown Rendering	Full GFM markdown in assistant responses with syntax highlighting
Conversation Export	Export any conversation as Markdown or JSON

Agents and Orchestration

Feature	Description
Custom Agents	Create AI personas with unique system prompts, skills, and behavior profiles
Agent Orchestration	Visual pipeline builder for chaining agents in multi-step workflows
Agent Import/Seed	Import agent configurations or seed from 46 built-in agent templates
Skills System	50+ modular skill files for specialized AI behaviors (coding, research, design, finance, etc.)
Self-Improving Agents	Agents that record experiences, reflect on outcomes, and evolve their prompts over time
Agent Experience Tracking	Execution scoring and performance metrics stored per agent
Autonomous Coding	Full Plan-to-Code-to-Test-to-Fix-to-Commit pipeline with rollback support
Swarm Coordination	Queen-led agent swarms with consensus-based decision making
SONA Self-Learning	Neural pattern recognition and trajectory learning for adaptive agent behavior

Memory and Knowledge

Feature	Description
Universal Context Memory	Cross-session context retention with HNSW vector-based semantic retrieval
Memory Linking	Connect related memories across conversations and agents
Context Compression	Automatic compression of long contexts to maintain relevance
Context Pruning	Intelligent pruning of stale or low-value memory entries
Knowledge Base	Document ingestion, chunking, embedding, and semantic search
Knowledge Graph	Entity-relationship mapping with graph traversal and visualization
Memory Statistics	Real-time stats on memory usage, linkage density, and recall accuracy

Artifact System (Claude-Artifacts Style)

Feature	Description
Auto-Detection Engine	Automatically detects 11 artifact types in AI responses
Live Streaming Preview	Right-side resizable preview panel with fullscreen mode
Multi-Tab Interface	Pin, favorite, and navigate recent artifacts
Document Export	Export to DOCX, PDF, XLSX, PPTX, Markdown, HTML, CSV
Shareable Links	Generate shareable links with view tracking
Version History	Full version history with rollback capability
Sandbox Preview	Isolated sandbox for React/HTML/Tailwind code rendering

Document Generation Engine

Format	Capabilities
DOCX	Reports, proposals, resumes with sections, bullets, tables, and styling
PDF	Styled paginated documents with embedded fonts and layout control
XLSX	Multi-sheet workbooks with headers, formatting, and formulas
PPTX	Presentation slides with bullets, layouts, and speaker notes
CSV / Markdown / HTML	Direct flat-file exports with automatic format detection

Scheduling and Automation

Feature	Description
Cron Scheduler	Schedule agents and tasks with full cron expression support
Cron Workers	Background worker processes for always-on agent execution
Self-Healing Tasks	Automatic detection and recovery of failed scheduled tasks
Scheduler Panel	Visual UI for managing, monitoring, and testing cron jobs

Enterprise and Security

Feature	Description
AI Defence System	Prompt injection detection, PII scanning, and threat level monitoring
Encryption Vault	AES-256-GCM at-rest encryption for sensitive data
Zero-Trust Federation	Cross-machine mutual TLS with Ed25519 key verification
Compliance Engine	Audit log management, policy enforcement, and automated compliance scanning
Security Vault	Secret scanning, exposure detection, and vulnerability tracking

Collaboration and Multi-Device

Feature	Description
LAN Network Access	Run the dashboard on your machine, access from any device on the network
Auto-Detected IP	Local IP automatically discovered with shareable URL
CORS Headers	Cross-origin access support for web integrations
Real-Time Collaboration	Shared sessions with synchronized editing and context
Mobile Companion	Monitor your dashboard from a mobile device with push notifications

---

8-Tier Power Tools System

Tier 1 — Game Changers

Tool	Description
Agent Orchestration	Visual multi-step pipeline builder for chaining agents, with approval gates, pause/resume, and status tracking
Autonomous Coding	Full software development loop — Plan, Code, Test, Fix, Commit — with iteration, rollback, and plan management
Model Router	Smart AI model switching based on task type, latency, and cost; per-conversation model selection with grouped dropdowns
Codebase Intelligence	Semantic code search, security vulnerability scanning, dependency analysis, and code indexing across your project
AI Pair Terminal	AI monitors your terminal output in real time, suggests commands, and detects errors proactively
Kanban Board	Task management board with AI agent assignment, card movement, and progress tracking
Architecture Mapper	Codebase visualization, dependency graphing, and architectural pattern detection

Tier 2 — Power Features

Tool	Description
Comms Hub	Unified messaging across WhatsApp, Telegram, Discord, and Slack with broadcast and status monitoring
Cross-Provider Consensus	Ensemble voting — send a prompt to multiple AI providers simultaneously and aggregate the best answer
Visual UI Builder	Screenshot-to-component pipeline; generate React/Next.js components from visual mockups with AI
Database Studio	AI-powered database management — connect, query, schema inspection, and migration execution

Tier 3 — Pro Features

Tool	Description
Deploy Pipeline	One-click deployment with environment management, rollback, and real-time log streaming
Security Vault	Secret scanning, exposure detection, compliance auditing, and security vulnerability tracking
Analytics and Insights	Token usage analytics, AI performance metrics, and event tracking with visualization
Plugin Marketplace	Community extension registry with install, seed, and management capabilities

Tier 4 — Differentiators

Tool	Description
Quick Actions	Right-click context AI actions for instant transformations on selected text or code
Voice-to-Code Pipeline	Speak natural language and have AI generate production code in real time
Git Intelligence	AI-powered commit messages, PR reviews, conflict analysis, and repository insights
Mobile Companion	Monitor your dashboard from a mobile device with push notifications
Web Bridges (WebBridge)	Free AI access via browser token extraction — auto-grab, validate, and use tokens from browser sessions

Tier 5 — New Generation

Tool	Description
AI Artifacts Studio	Generate and preview documents, spreadsheets, slides, and visual canvases with streaming AI output
LAN Network Access	Multi-device WiFi access — run the dashboard on your machine, access from any device on the network
Visual Canvas	Fabric-like design editor for drag-and-drop layout creation and visual editing

Tier 6 — Next-Gen OS

Tool	Description
Universal Memory	Cross-session context with HNSW vector-based semantic recall and intelligent linking
Cron Scheduler	Always-on agents and scheduled task execution with worker management and self-healing
Deep Research	Multi-source research with citations, source verification, hallucination detection, and export
Issue-to-Deploy Pipeline	End-to-end pipeline from GitHub issue through code generation to deployed solution
Self-Improving Agents	Prompt optimization through experience recording, reflection, evolution, and recommendation cycles
Hybrid Router	Local-to-cloud provider racing — benchmark providers on latency, cost, and quality; auto-select the best
Live Sandbox	Instant application preview with isolated execution environments and deployment
MCP Hub	Model Context Protocol server registry, tool discovery, execution, and management across transports
Compliance Engine	Audit log management, compliance policy enforcement, and automated scanning
Collaboration	Real-time multiplayer sessions with shared context and synchronized editing

Tier 7 — Preconfigured OS

Tool	Description
Prebuilt Agents	One-click activation of 46 specialized agents across 10 categories
Pipeline Templates	7 production-ready pipeline configurations for common workflows

Tier 8 — Enterprise AI Engine

Tool	Description
Swarm Coordination	Queen-led agent swarms with consensus-based decision making and task distribution
SONA Self-Learning	Neural pattern recognition, trajectory learning, and adaptive behavior optimization
Zero-Trust Federation	Cross-machine mutual TLS with Ed25519 key verification and PII scanning
Encryption Vault	AES-256-GCM at-rest encryption for sensitive data and credentials
AI Defence System	Prompt injection detection, PII scanning, threat level monitoring, and safety guardrails
Cost Tracker	Real-time token cost tracking, budget alerts, and provider cost comparison
Goal Planner	Strategic goal decomposition with action plans, progress tracking, and milestones
Knowledge Graph	Entity-relationship mapping with graph traversal, visualization, and semantic queries
Background Workers	Always-on worker processes for scheduled tasks, health checks, and autonomous operations
Verification Engine	Cryptographic verification, witness attestation, and integrity proof generation

---

Platform Overview

Category	Detail
Framework	Next.js 15 (App Router) + React 19
Language	TypeScript 5
Styling	Tailwind CSS + shadcn/ui + Radix UI
Database	Prisma ORM with SQLite
State Management	Zustand
API Endpoints	70+ route files (100+ HTTP endpoints)
Database Models	40 Prisma models
UI Components	38 custom components + 37 shadcn/ui primitives
Enhancement Panels	44 feature panels across 8 tiers
Core Libraries	30 internal modules
Skills	50+ built-in skill modules
Prebuilt Agents	46 agents across 10 categories
Pipelines	7 production-ready pipeline templates

---

Architecture

INFOHASCLAWHUB/
├── src/
│   ├── app/
│   │   ├── api/                        # 70+ API route files
│   │   │   ├── agents/                 # Agent CRUD, run, import, seed
│   │   │   ├── artifacts/              # Artifact generation and sharing
│   │   │   ├── bots/                   # Comms hub connections and broadcast
│   │   │   ├── browser/                # Token extraction and Playwright
│   │   │   ├── codebase/               # Code indexing, search, security
│   │   │   ├── coding-sessions/        # Autonomous coding pipeline
│   │   │   ├── collab/                 # Collaboration sessions
│   │   │   ├── compliance/             # Compliance engine and policies
│   │   │   ├── conversations/          # Chat history and messages
│   │   │   ├── cron/                   # Scheduling, workers, self-heal
│   │   │   ├── db-studio/              # Database connections and queries
│   │   │   ├── deploy/                 # Deploy pipeline and rollback
│   │   │   ├── gemini/                 # Gemini CLI chat
│   │   │   ├── git/                    # Git intelligence
│   │   │   ├── inference/              # Streaming inference
│   │   │   ├── issue-pipeline/         # Issue to deploy
│   │   │   ├── kanban/                 # Kanban boards and cards
│   │   │   ├── knowledge/              # Knowledge base documents
│   │   │   ├── mcp/                    # MCP server management
│   │   │   ├── memories/               # Memory CRUD
│   │   │   ├── memory/                 # Context memory engine
│   │   │   ├── model-routes/           # Model routing
│   │   │   ├── orchestrator/           # Agent orchestration
│   │   │   ├── pipelines/              # Pipeline execution
│   │   │   ├── plugins/                # Plugin marketplace
│   │   │   ├── provider-router/        # Hybrid provider racing
│   │   │   ├── providers/              # Provider management
│   │   │   ├── research/               # Deep research with citations
│   │   │   ├── sandbox/                # Live sandbox environments
│   │   │   ├── scheduler/              # Task scheduling
│   │   │   ├── security/               # Security scanning and audit
│   │   │   ├── self-improving/         # Agent self-improvement loop
│   │   │   ├── ui-builder/             # Visual UI builder
│   │   │   ├── v1/                     # OpenAI-compatible gateway
│   │   │   ├── whatsapp/               # WhatsApp integration
│   │   │   ├── telegram/               # Telegram bot integration
│   │   │   └── discord/                # Discord bot integration
│   │   ├── globals.css
│   │   ├── layout.tsx
│   │   └── page.tsx                    # Main dashboard
│   ├── components/
│   │   ├── enhancements/               # 44 feature panels (8 tiers)
│   │   ├── ui/                         # 37 shadcn/ui primitives
│   │   ├── ChatWindow.tsx              # Main chat interface
│   │   ├── ChatInput.tsx               # Message input with attachments
│   │   ├── ChatSidebar.tsx             # Conversation list sidebar
│   │   ├── TopBar.tsx                  # Navigation bar and power tools
│   │   ├── SettingsPanel.tsx           # Application settings
│   │   ├── CommandPalette.tsx          # Ctrl+K command palette
│   │   └── ...                         # Additional components
│   └── lib/
│       ├── stores.ts                   # Zustand global state
│       ├── db.ts                       # Prisma client singleton
│       ├── inference.ts                # AI inference engine
│       ├── openai-gateway.ts           # OpenAI-compatible gateway
│       ├── provider-router.ts          # Hybrid provider routing
│       ├── orchestrator.ts             # Agent orchestration engine
│       ├── agent-runner.ts             # Agent execution runtime
│       ├── universal-memory.ts         # Context memory engine
│       ├── cron-engine.ts              # Cron scheduling engine
│       ├── research-engine.ts          # Deep research with citations
│       ├── self-improving.ts           # Agent self-improvement
│       └── ...                         # Additional modules
├── prisma/
│   └── schema.prisma                   # 40 database models
├── skills/                             # 50+ skill modules
└── db/                                 # SQLite database (auto-created)

---

API Reference

Chat and Inference

Route	Methods	Description
/api/run	POST	Execute AI inference
/api/inference/stream	POST	Stream AI inference response
/api/generate	POST	Generate AI completion
/api/v1/chat/completions	POST	OpenAI-compatible chat endpoint
/api/v1/models	GET	OpenAI-compatible model listing
/api/gemini/chat	POST	Gemini CLI chat interface
/api/consensus	POST	Multi-model consensus voting

Conversations and Messages

Route	Methods	Description
/api/conversations	GET, POST	List and create conversations
/api/conversations/[id]	GET, PATCH, DELETE	Get, update, or delete a conversation
/api/conversations/[id]/messages	GET, POST	Get or add messages
/api/messages/[id]	PATCH, DELETE	Update or delete a message

Agents

Route	Methods	Description
/api/agents	GET, POST	List and create agents
/api/agents/[id]	GET, PATCH, DELETE	Agent CRUD operations
/api/agents/run	POST	Execute an agent
/api/agents/import	POST	Import agent configuration
/api/agents/seed	POST	Seed built-in agents

Memory and Knowledge

Route	Methods	Description
/api/memories	GET, POST	List and create memories
/api/memory/context	GET, POST	Get or set active context
/api/memory/context/compress	POST	Compress context window
/api/memory/context/prune	POST	Prune stale entries
/api/memory/context/stats	GET	Memory statistics
/api/knowledge/documents	GET, POST	Knowledge base documents
/api/knowledge/search	POST	Semantic knowledge search

Orchestration and Pipelines

Route	Methods	Description
/api/pipelines	GET, POST	List and create agent pipelines
/api/pipelines/[id]/run	POST	Execute a pipeline
/api/pipelines/[id]/approve	POST	Approve a pipeline step
/api/orchestrator/start	POST	Start orchestration session

Scheduling and Automation

Route	Methods	Description
/api/cron/tasks	GET, POST	List and create cron tasks
/api/cron/workers	GET, POST	Worker management
/api/cron/execute	POST	Execute a cron task immediately
/api/cron/self-heal	POST	Self-heal failed tasks

Comms Hub

Route	Methods	Description
/api/bots/connections	GET, POST	List and create bot connections
/api/bots/connections/[id]/connect	POST	Establish connection
/api/bots/connections/[id]/send	POST	Send message
/api/bots/broadcast	POST	Broadcast to all channels

Security and Compliance

Route	Methods	Description
/api/security/compliance	GET	Compliance status
/api/security/audit	GET, POST	Audit logs
/api/security/secrets	GET, POST	Secret scanning
/api/compliance/policies	GET, POST	Policy management

System and Utilities

Route	Methods	Description
/api/health	GET	Health check
/api/doctor	GET	Diagnostic report
/api/system/monitor	GET	System resource monitoring
/api/settings	GET, PATCH	Application settings
/api/skills	GET, POST	Skill management
/api/prompts	GET, POST	Prompt template library
/api/search	POST	Global search

---

Keyboard Shortcuts

Shortcut	Action
Enter	Send message
Shift+Enter	New line
Ctrl+N	New chat
Ctrl+B	Toggle sidebar
Ctrl+K	Open command palette
Ctrl+,	Open settings
Escape	Close dialogs

---

Environment Variables

Create a .env file in the project root (defaults are included):

DATABASE_URL="file:./db/clawhub.db"

Additional provider keys can be configured through the Settings > Providers UI — no environment variables needed for most setups.

---

Troubleshooting

Issue	Solution
gemini is not recognized	Run npm install -g @google/gemini-cli, restart terminal, then gemini --version
Gemini auth fails on WSL	Run gemini in WSL terminal and complete the browser OAuth flow
Database errors	Run npx prisma db push --accept-data-loss to reset the database
Port 3000 already in use	lsof -i :3000 then kill -9 <PID> (Linux/WSL) or netstat -ano | findstr :3000 then taskkill /PID <PID> /F (Windows)
Provider connection failures	Check API keys in Settings > Providers and verify base URLs
Dashboard too large on small screen	The layout auto-scales for screens 1280px and below. Ensure browser zoom is at 100%
WSL can't access localhost:3000	Use hostname -I to get WSL IP, then access http://<WSL-IP>:3000 from Windows

---

License

MIT — Free for personal and commercial use.