Features
Everything your AI agent can do — out of the box
Core Agent Runtime
The foundation — built in Rust for speed and reliability
Isolated Cloud VM
Each agent runs in its own secure cloud sandbox with dedicated CPU, memory, and persistent filesystem. Full Linux environment — not a shared container.
Rust-Powered Runtime
LunoGen is built 100% in Rust — <10ms cold start, 8.8MB binary, <5MB RAM idle. 10x more efficient than Node.js-based alternatives.
Persistent Memory
Vector embeddings with semantic search. Your agent remembers every conversation, fact, and preference across sessions. Supports SQLite, PostgreSQL, Qdrant, and 5 more backends.
Full Shell & Code Execution
Execute any shell command, run Python, Node.js, or Bash scripts. Install packages, process data, deploy code — your agent has a real development environment.
Cron Scheduling & Reminders
Schedule recurring tasks with standard cron syntax — daily briefs, weekly reports, follow-up reminders. Your agent works autonomously while you sleep.
Auto-Healing Runtime
Self-restarts on crash, config versioning with one-click rollback, automatic settings preservation across restarts. Zero-maintenance uptime.
AI Models & Intelligence
20+ providers, smart routing, and model fallbacks
20+ AI Providers, Latest Models
Claude Opus 4.6, Claude Sonnet 4.6, GPT-5.4, GPT-5.2 Mini, Gemini 3.2 Flash, Gemini 3 Pro, DeepSeek R2, Llama 4, Mistral Large, Command R+, Qwen 3 and more. Switch models on a live agent — no restart needed.
Model Fallbacks & Chains
Configure backup models that automatically activate when your primary is down or rate-limited. Chain multiple fallbacks for zero-downtime reliability.
Smart Query Routing
Automatically route queries to the best model — send coding questions to Claude, simple chat to a cheaper model, research to GPT. Per-tool model assignment supported.
Embedding & Vector Search
Semantic memory powered by latest embedding models — OpenAI text-embedding-3-large, Gemini embedding, or any OpenRouter-compatible model. Hybrid keyword + vector retrieval.
Voice Message Processing
Transcribes voice notes from WhatsApp and other channels using Groq Whisper (fast) or OpenAI Whisper. Responds to voice as naturally as text.
Autonomy Levels
Three modes: Read-Only (observe only), Supervised (asks before acting), Full Autonomy (acts independently). Control how much freedom your agent has.
Channels & Messaging
WhatsApp, Telegram, and API access
WhatsApp (QR Code)
Scan a QR code — just like WhatsApp Web. No API keys, no business account. Works with your personal WhatsApp. Live in under a minute.
WhatsApp Cloud API
Official Meta Business API with webhook-verified delivery. Handle unlimited conversations at scale. Perfect for businesses.
Telegram
Native Telegram bot integration via BotFather. Commands, inline buttons, media sharing, typing indicators, and sender allowlists.
Webhook & WebSocket
Every agent gets an HTTPS webhook URL and WSS streaming endpoint. POST JSON for request/response, or stream tokens in real-time.
Media Generation
Create images, videos, audio, and documents
Image Generation
Generate images via OpenAI GPT-Image, DALL-E, Replicate, Stability AI, Together AI, and Ideogram. Multiple sizes, styles, and quality options.
Video Generation
Create videos with Google Veo 3.1, Runway Gen-3, and Kling. Configurable duration, aspect ratio, and style. Polling-based generation with progress tracking.
Audio & Voice Generation
Text-to-speech via OpenAI TTS, ElevenLabs, Cartesia, Google Gemini TTS, and Suno for music. Multiple voices, formats (MP3, WAV, FLAC), and languages.
PDF & Document Creation
Generate PDFs with fpdf2, create data reports with matplotlib charts, process and create Excel spreadsheets, and build presentations.
Data Analysis
Analyze CSV files, create charts and visualizations with pandas + matplotlib, process spreadsheets with openpyxl. Full Python data science stack available.
Web Research
Search the web via DuckDuckGo or Brave, fetch and read any URL, extract content from PDFs and web pages. Compile research reports autonomously.
Plugins, MCP & Integrations
Extend your agent with 50+ services and custom tools
Plugin Library
Activate integrations with one click — GitHub, Slack, Notion, Stripe, Airtable, Linear, Shopify, Google Workspace, and more. Credentials auto-injected.
MCP Server Support
Connect Model Context Protocol servers via stdio, HTTP, or SSE transports. Use official MCP servers from Anthropic, Stripe, GitHub, and the community.
Composio (1000+ OAuth Apps)
Connect Gmail, Salesforce, Jira, HubSpot, and 1000+ apps via Composio OAuth. Your agent reads emails, creates tickets, sends invoices — no API keys needed.
Custom Skills
Create reusable skills in markdown — teach your agent new abilities with natural language instructions. Full-screen editor with live preview.
80+ Built-in Tools
Shell, browser, file ops, web fetch, memory, cron, image gen, video gen, audio gen, git, delegation, composio, notifications, and more. Enable or disable per-agent.
Multi-Agent Teams
Link agents together — delegate tasks, spawn sub-agents, coordinate workflows. Each agent has a role (researcher, coder, reviewer) and communicates via IPC.
Developer Experience
APIs, SDKs, and embeddable components
REST API
Full API to create, manage, and control agents programmatically. Manage tools, skills, plugins, MCP servers, and config versions via API key authentication.
WebSocket Streaming
Real-time token streaming via WSS. Build live chat experiences with tool call progress, typing indicators, and structured response events.
Embeddable Copilot Widget
Add an AI chat widget to any website with one script tag. Customizable theme (light/dark, accent color), CORS protection, rate limiting, and session persistence.
Terminal Access
Full WebSocket terminal to your agent's VM — right from the dashboard. Run commands, inspect files, debug issues in real-time.
Postgres + pgvector Sidecar
Optional PostgreSQL 16 with pgvector extension. Full SQL database for structured data, vector similarity search, and shared state. Auto-provisioned.
Headless Browser (Playwright)
Full Chromium browser inside the VM via Playwright. Navigate pages, fill forms, take screenshots, scrape data, and automate any web application.
Security & Operations
Enterprise-grade security, monitoring, and cost controls
VM Isolation & Sandboxing
Each agent runs in a dedicated container with dropped capabilities, seccomp filters, noexec workspace, PID limits, and blocked cloud metadata endpoints. Docker, gVisor, or Firecracker backends.
Prompt Guard & Leak Detection
Detects prompt injection attempts and scans LLM responses for accidentally-exposed secrets (API keys, tokens). Auto-redaction before delivery.
Emergency Stop
Multi-level kill switches — freeze specific tools, block network domains, or halt the entire agent instantly. Resume when ready.
Encrypted Credential Vault
All API keys and tokens encrypted with AES-256-GCM at rest. Per-agent credential overrides. Secrets never logged or exposed in responses.
2FA & Pairing Auth
Optional TOTP two-factor authentication. Gateway pairing with one-time codes and bearer tokens. Brute-force protection with lockout.
Policy Engine & RBAC
Fine-grained tool, domain, and network policies. Role-based access control. Audit logging for every action. Configurable per autonomy level.
Cost Tracking & Limits
Set daily and monthly spending limits with configurable warning thresholds. Real-time cost dashboard per agent. Auto-pause on budget exceeded.
Observability & Tracing
Full request tracing, structured runtime logs, health monitoring, and diagnostics. Agent Doctor for one-click troubleshooting.
Config Versioning & Snapshots
Automatic config snapshots before every restart and settings change. Full version history with one-click restore. Workspace backup and recovery.
Hot Config Reload
Change model, provider, tools, personality, and settings on a live agent — no restart needed. Soft restart reloads config in 2 seconds.
Ready to deploy your agent?
All features included. Bring your own AI key. Deploy in under a minute.